Political dynasties and personal political machines

Diarmuid Scully: 8904316

Supervisors: Dr Maura Adshead and Dr Rory Costello

1

Abstract

This thesis proposes an -based explanation for the observable variations in the rate of family following in the parliaments of mature, developed, democratic states. Electoral systems that foster intra-party competition for votes from the general public are held to incentivise the creation of personal political machines. Such machines are heritable and it is this inheritance that conveys significant advantages to the relatives of former MPs who follow them into elected office. A theory of dynasty formation and a new ranking of electoral systems according to the degree to which they incentivise the creation and maintenance of personal political machines are developed. These are tested against recent in seven countries and against the last eleven elections in Ireland. The results support the thesis and lead to a series of recommendations for electoral system reform.

2

Table of Contents

List of Graphs and Tables Page 4

Chapter 1: Introduction and Outline of Study Page 7

Chapter 2: Literature Review Page 15

Chapter 3: Case Selection and Methodology Page 53

Chapter 4: Results Page 84

Chapter 5: Longitudinal Case Study: Ireland Page 152

Chapter 6: Summary and Conclusion Page 196

References Page 215

3

List of Graphs and Tables

Table 1: Electoral systems used in Mature, Developed, Democratic States page 58

Table 2: Electoral systems ranked by Incentive to Create and Maintain a Personal Political Machine: page 66 Table 3: Countries in Study by Electoral System page 71

Table 4: Countries in study by Electoral System Features page 72

Graph 1: Rate of Family Following by Electoral System page 85

Graph 2: Rate of Family Following with New Zealand as two systems page 86

Table 5: Rate of Family Following by Country page 87

Table 6: Rate of Family Following by Electoral System Features page 88

Graph 3: Number of Family followers per country page 88

Table 7: Results Summary by Country page 89

Table 8: Rate of Family Following in Australian State page 97

Graph 4: Effect of Personalised on rate of Family Following page 110

Table 9: Personalised Voting z Test page 111 Graph 5: Effect of Intra-party Competition on rate of Family Following page 112 Table 10: Intra-party Competition z Test page 112 Graph 6: Effect of Partial and Full Intra-party Competition page 113 Graph 7: Effect of Preference vote seeking from other parties page 114

Graph 8: Preference vote seeking from other parties with and without Intra-party Competition page 115 Graph 9: Intra-party Competition and Same-Constituency Inheritance page 117 Graph 10: Effect of Partial,Full IPC and Preference vote seeking on rate of Same-Constituency Inheritance page 119 Graph 11: Effect of Intra-party competition on rate of Direct Inheritance page 120 Graph 12: Direct Inheritance by intra-party competition separating Belgium page 121 Graph 13: % of Family Followers and Traditionally Successful Parties by intra-party competition (excluding USA and Canada) page 124 Graph 14: Effect of Intra-party Competition on proportion of Cabinet who are Family Followers page 126

4

Graph 15: % of MPs who are current or former Cabinet Ministers by intra-party competition page 129 Graph 16: MPs holding Prior Local/Regional Elected Office page 130 Table 11: Family Following and Corruption Perceptions Index page 132

Graph 17: Rent Seeking and Family Following page 133

Table 12: Family following and relative development page 134

Graph 18: Family following and UNHDI page 135

Graph 19: Family following in Urban and Rural constituencies page 136

Table 13: Rural z Test page 138

Graph 20: Rates of Family Following and Proportion of MPs from Rural Constituencies page 138

Graph 21: Family Following and Graduates page 139

Table 14: Graduates z Test page 140

Graph 22: Socially Conservative MPs and Family Following page 141

Graph 23: Women as % of all MPs by Country page 144

Graph 24: Family Following by Gender and Intra-Party Competition page 145

Table 15: Multivariate Analysis page 146

Table 16: Probability Table Electoral Systems- controlling for all other factors page 147

Graph 25: Rate of Family Following controlling for other factors with 95% Confidence Intervals page 149

Graph 26: Rate of Family Following, Ireland, 1981 – 2016 page 172

Graph 27: Same-Constituency Inheritance 1981 -2016 page 174

Graph 28: Direct v Delayed Inheritance page 175

Graph 29: TDs from Traditionally Successful parties page 176

Graph 30: Family Followers and Traditionally Successful Parties page 177

Graph 31: Family Seats Dáil and Cabinet 1981 – 2016 page 180

Graph 32: Prior Experience in Local page 181

Graph 33: Family Seats by Region page 184

5

Graph 34: Graduates Ireland 1981 - 2016 page 186

Graph 35: Family following by Party page 189

Graph 36: Average age at 1st 1981 – 2016 page 191

Graph 37: Women in the Dáil page 192

6

Chapter 1: Introduction and Outline of Study

1.1 Introduction:

Political dynasties are a common feature in many established . Four of the nine men who have held the office of of Ireland since 1970 - , Garret

Fitzgerald, and - have been the sons of former TDs, while a fifth,

Charles Haughey, was the son-in-law of a former Taoiseach. Though the current Taoiseach,

Leo Varadkar, is not a member of a political dynasty, the party rival he defeated to secure that office, is.

By contrast, when George W Bush was elected President of the United States he became just the second son to succeed his father to that office in over 200 years of US history and only the third to be a member of a Presidential dynasty – the others being Presidents John

Quincy Adams, son of John Adams, and Benjamin Harrison, grandson of William Henry

Harrison (US Congress 2016).

The man Bush defeated in 2000, Al Gore, however, was himself a family follower at a

Congressional level, succeeding his father Al Gore Snr to seats in both the US House of

Representatives and Senate. In the most recent US election two prominent dynastic candidates ran, neither of whom was elected. Jeb Bush, the son of one President and brother of another, unsuccessfully sought the Republican nomination. Hillary Clinton, spouse of former President Bill Clinton, did secure the Democratic nomination, and even a

7 plurality of the public vote in the general election, but was ultimately unsuccessful. Dynastic succession in the US is more common at congressional level, where, since the 1960s, some

6% of members of the House of Representatives have been members of political dynasties

(Dal Bó et al. 2009). The longest serving US dynasty – the Frelinghuysens– have represented the state of New Jersey either in the House, the Senate or the Continental Congress in unbroken line since 1753. The current incumbent, Rodney P Frelinghuysen, was returned to

Congress in November 2016 with 58% of the vote (US Congress 2016).

Family following runs at a lower rate in Canada, where, over the last four decades, some 3% of members of the House of Commons have been closely related to former MPs (Godwin

2013). Most notably, the current Prime Minister Justin Trudeau is the son of former Prime

Minister Pierre Trudeau, but he is not alone. Nancy Ruth MP of the Progressive

Conservative Party is the daughter and granddaughter of MPs. Maxime Bernier, Daniel

Blaikie, Geoff Regan and Mark Strahl all hold the same seats in Parliament that were once held by their fathers, while husband and wife Diane and Doug Finlay served alongside one another until Doug’s death in 2013 (Parliament of Canada 2015).

In the UK, former MP Tony Benn, who gave up his right to a hereditary seat in the House of

Lords, was followed into the House of Commons by his son Hilary. This is not unusual.

Thirty four MPs in the parliament elected in 2010 were close family relations of current or previous members, among them the then leader of the Ed Milliband, brother of David and the husband and wife team of Ed Balls and Yvette Cooper, while on the

8

Conservative benches, former cabinet members Douglas Hurd and John Selwyn Gummer were followed respectively by their sons Nick Hurd and Ben Gummer (Dods 2010).

In the Netherlands, Pieter Duisenberg, son of former MP, Minister and ECB Director Wim

Dusienberg sits in the National Parliament along with Pieter Heerma, son of former MP and

Minister Enneüs Heerma (Parliament of Netherlands 2016).

In the current Prime Minister Shinzō Abe is the son of a former foreign Minister and the grandson of a former Prime Minister. Abe’s first term as head of government ended with his resignation in 2007. His replacement on that occasion was – another family follower and the son of a former Prime Minister. Until recently in excess of 30% of

Japanese MPs were similarly closely related to former office holders (Kato S 2009; Smith

2012) though this began to decline slowly following a change in the electoral system from

Single Non-Transferable Vote (SNTV) to Mixed Member Plurality (MMP) in 1993 (Smith

2012; Doyle et al. 2015).

In Malta former Minister Carmelo Mifsud Bonnici is the son and grandson of MPs, as is the

Deputy Leader of the Nationalist Party Beppe Fenech Adami. The son of former Prime

Minister Yitshak Shamir sits in the Israeli Knesset alongside the son of former President

Chaim Herzog and the daughter of former Foreign minister David Levy (Knesset 2014). Even the rise of the far right in French has a familial flavour as succeeded her father Jean Marie as leader of the Front National and as a Member of the European

9

Parliament and followed in his footsteps by contesting the Presidency. Her niece Marion

Maréchal le Pen became France’s youngest MP and both women topped the polls in the first round of last year’s French regional elections.

While family following occurs cross-nationally and across the political spectrum the rate of following between countries varies quite considerably from relative lows of 3% in Canada

(Parliament of Canada 2015) and 4% in Australia (Lumb 2012) to 30% in Japan (Kato S 2009;

Smith 2012) and 65% in the (Riedinger 1995). It is this variation in the rate of family following between countries that is the subject of this study.

A number of explanations have been advanced in the literature and these will be looked at more closely in Chapter 2, but in general higher rates of family following are seen as a negative indicator for the political health of a society. The existence of persistently higher rates of family following in some countries has been put down to a number of factors including underdevelopment (Putnam 1976; Riedinger 1995), corruption and state capture

(Acemoglu and Robinson 2006: 26; Camp 1982: 850-851; Riedinger 1995: 209), the openness of the electoral system (Van Liefferinge and Steyvers 2007; Kato S 2009; Dal Bó et al. 2009; Smith 2012; Van Coppenolle 2014) and the degree of local control of the nominating process (Gallagher 1988; Smith 2012; Doyle et al. 2015).

To control for underdevelopment and corruption, this study focuses on mature, developed, democratic states with low levels of perceived corruption. The methodology for the case

10 selection is outlined in Chapter 3. Even within the group of states selected the rate of family following varies quite considerably from less than 2% in the Netherlands to almost

20% in Ireland.

Chapter 2 outlines a Theory of Dynasty Formation which argues that a key explanatory variable for the observable variations in the rate of family following among mature, developed, democratic states with low levels of perceived corruption, is the degree to which the electoral system incentivises candidates to create and maintain personal political machines. Personal political machines are defined as a network of clients, canvassers, fundraisers, election and constituency workers whose primary loyalty is to the candidate and not to the party. These machines are heritable and, as is outlined in Chapter 2, their existence increases both the supply of and demand for dynastic candidates. This variable is similar to the concept of Electoral System openness but differs in that it is candidate focused rather than voter focused. This difference, and its implications, will be discussed further in

Chapter 3.

If supported, this theory has considerable implications for electoral system choice. There has in recent years been a growing trend of states seeking to combat public disengagement from politics by reforming their electoral systems to increase the level of personalisation

(Renwick and Pilet 2016). While on the surface these reforms appear to offer a greater degree of voter choice, the theory advanced in this thesis suggests that, if implemented in

11 the wrong way, they may instead have the perverse effect of reducing voter choice by raising barriers to participation and entrenching an oligarchic political class.

1.2 Organisation of Thesis:

The opening chapter introduces the concept of family following, puts forward the central proposition of this thesis and briefly outlines the approach being taken. Chapter 2 reviews the existing literature on family following, posits a Theory of Dynasty Formation and develops a series of testable hypotheses and additional oberservable implications from this theory. It also outlines a number of alternative explanations for the existence of dynasties and proposes how these may be tested. Chapter 3 outlines the case selection and methodology.

Chapter 4 analyses the results of the study of family following in seven mature, developed democratic states on both a national basis and on a comparative cross-national legislator level and tests these against the hypotheses, additional observable implications and alternative hypotheses posited in Chapter 3.

Chapter 5 is a longitudinal case study of family following in a single country, Ireland, from

1981 – 2016. This longitudinal study allows for further testing of some of the hypotheses, additional observable implications and alternative hypotheses over time. Chapter 6 concludes and points towards further research.

12

1.3 Definitions

A number of terms are used throughout this thesis and have the following meaning unless otherwise specified:

Family following is held to occur whenever an MP is elected to parliament who is a close relative of a current or former member of the same or higher parliamentary body. A close relative is a relation through blood or marriage of two degrees or less i.e. children, grandchildren, parents, grandparents, siblings, spouses, nieces, nephews, grandnieces, grandnephews, and equivalent in-laws.

A is a collective term used to refer to a number of current or former office holders who are closely related while a family seat is any seat in a national parliament held by a member of a political family.

Family followers can inherit a family seat in one of three ways. Direct inheritance occurs when the parliamentarian is elected to the same seat held by their relative in the election immediately following that relatives death, defeat or retirement. Delayed inheritance occurs when the parliamentarian is elected to the same seat held by their relative at an election after that immediately following the death, defeat or retirement of their relative.

Extra-constituency following occurs when a parliamentarian is elected to a different seat to their relative, but to one in the same house of parliament.

13

As defined above, a personal political machine is a network of clients, canvassers, fundraisers, election and constituency workers whose primary loyalty is to the candidate and not to the party.

This study concerns family following in mature, developed democratic states which are defined as states which have been democracies for at least 45 years and enjoy both political freedom and a high human development index score (see detail under case selection in

Chapter 2).

Intra-party competition is defined as competition between members of the same for votes from the public either at general election or in the case of the United States in an open primary. Preference vote seeking from other parties refers to that element of some electoral systems that allows a voter to give a preference vote to a candidate of a party other than that of the recipient of their 1st vote.

14

Chapter 2: Literature Review

2.1 Dynasties in Democracies

Family following has been a feature of representative since its inception (Michels

1962) and has long been controversial. The issue of family following in mature, developed, democratic states is of interest in its own right and has been the subject of a number of single country studies in the United States (Laband and Lentz 1985; Hess 1997; Dal Bó et al.

2009) Japan (Reed 1994; Kato H 2009; Kato S 2009), Ireland (Smith and Martin 2017),

Belgium (Van Liefferinge and Steyvers 2007; Van Coppenolle 2014), Australia (Lumb 2012),

Norway (Fiva and Smith 2016) and Italy (Daniele and Geys, 2014) and two comparative cross-national studies (Smith 2012; Doyle et al. 2015). One recent US study (Dal Bó et al.

2009) has found that politics is the most heritable of all professions ahead even of farming and self-employed proprietorship, while another found that it is in fact the third most heritable, but only if billionaire and reality TV star are considered as professions (Stephens-

Davidowitz 2015). That the children of elected are more likely to follow in their parents’ footsteps and become elected politicians themselves than the children of farmers are to take up farming or the children of small shop owners are to take on the family business (Dal Bó et al. 2009) raises important questions about the functioning of democracy.

As Marsh (1987) points out, the question of how candidates for election are recruited is of critical importance to the democratic process. Family following is clearly a significant factor in this process and is therefore worthy of broader study.

15

Some see family following as the embodiment of inequality leading to the entrenchment of vested interests and inferior making (Kato S 2009; Kato H 2009; Querubin 2012;

Mendoza at al 2012; Daniele and Geys, 2014; Stephens-Davidowitz 2015; Economist 2015).

It is alleged that family followers are less qualified than non-followers (Feinstein 2010;

Daniele and Geys, 2014), and may even be more prone to corruption (Smith 2012; Daniele and Geys, 2014)

Evidence from the business world strongly supports this negative interpretation. A wide range of independent studies across several countries have found that companies where the

CEO is closely related to their predecessor perform less well, in terms of operating profit and overall shareholder value, that those led by a non-family member (Bennedsen et al.

2007; Morck et al. 2000; Daniele and Geys, 2014; Pérez-González 2006). A contrary point is made by Doyle et al. (2015) who argue that family owned businesses out-perform their publicly owned counterparts because they take a longer term view and invest for the future.

Much of the evidence supporting this favourable view of family owned firms comes from

Japan and is problematic due to the widespread Japanese custom of adult adoption

(Economist 2012). When Japan is excluded, family following in publicly listed companies is strongly correlated with poorer performance. Some 90% of Japanese adoptions, comprising many tens of thousands of cases a year, are of adult men in their twenties and thirties, primarily into families that own businesses (Economist 2015). These adoptees are rarely orphans, but rather successful managers often with families of their own. Upon adoption they take on the name and family history of their new clan abandoning their previous identity. For example the car company Suzuki is family owned, but it’s current CEO Osamu

16

Suzuki is the fourth adopted person to hold that position and has recently announced that he will be passing the company on to a fifth adopted son, by-passing his own biological heir.

Considerable evidence suggests that companies with adopted CEOs significantly outperfom those run by blood relatives (Mehrota el al 2013). The exact impact of this custom of adult adoption on the rate of family following in Japanese politics is not something that has been addressed specifically in the literature though Smith (2012) does make reference to the occurrence of such adoptions.

Hess (1997) also takes a benign view of family following arguing that because family followers are drawn from the elite of society, its “best butter”, that they have a natural advantage over non-followers, principally in terms of education. He even goes so far as to argue that there may be an as yet undiscovered genetic component to this advantage.

Even if true - and Hess offers no evidence in support of this concept of family followers as electoral übermensch - the implications for democracy of rule by a self perpetuating genetic elite would still be considerable and troubling.

Smith and Martin (2017) observe that in Ireland, family followers have generally enjoyed higher levels of education than non-followers. They further observe that cabinet ministers form an elite within the elite, with the relatives of former ministers being significantly more likely than either other family followers, or TDs who are not family followers, to become ministers themselves. While not dismissing the possibility that these advantages are the result of superior education, Smith and Martin (2017) propose that they are more likely the

17 result of an inherited “information advantage”, namely that the successors of backbench

TDs have learned from their predecessors how to gain and retain a parliamentary seat, but not how to advance to cabinet, while the successors of cabinet ministers have an inherited

“information advantage” in achieving both.

Others have argued, that because they are more likely to possess an independent power base, family followers are less reliant on their parties and therefore more independent minded than non-followers and are more receptive to the demands of voters (Laband and

Lentz 1985; Kato S 2009). Certainly, whatever the cause, a clear voter preference appears to exist for family followers over non-followers (Feinstein 2010; Van Coppenolle 2014). The argument has been made that such a preference may be in voters own self-interest as politicians who seek to establish a dynasty are more likely to continue to engage in constituency work even after they have decided not to stand for re-election (Laband and

Lentz 1985). There is also some evidence that dynastic politicians are more likely to vote against their party and in favour of local interests (Dal Bó et al. 2009; Reed 1994; Kato S

2009) though while this might benefit the voters in a particular constituency, it may be to the detriment of the electorate as a whole (Kato S 2009).

The oldest explanation offered in the literature for the existence of family following is that it is evidence of what Michels termed “the iron law of oligarchy” (Michels 1962[1911]) namely that elected representatives constitute a power elite that self-perpetuates (Mosca 1896). A more modern expression of this phenomenon is the concept of state capture (Camp 1982;

18

Acemoglu and Robinson 2006). When Camp (1982) looked at Mexican politics he found that the proportion of elected politicians from political families had remained essentially stable over the previous five decades at between 29% and 30%. These were the same families that dominated the country's business life and the faculty at the National University. Acemoglu and Robinson (2006) use the term “captured democracy” to describe how elites have used their economic power to dominate politics in Latin America, and Liberia to the present day and how they did so in the southern states of the US prior to 1960. More recent evidence comes from the Philippines where although the 1987 Constitution directed the state to outlaw political dynasties, the dynastically dominated Congress – in 1994 two thirds of

Congress members and almost all Senators were close relations of former parliamentarians

(Riedinger 1995) - simply refused to pass the necessary legislation. Two decades later it still hadn’t been passed (Queberin 2012). State capture is characterised by rent seeking behaviour on the part of politicians and is generally associated with imperfect democracies where political and economic freedom are lacking (Carson 2007; Leeson 2008; Queberin

2012).

The state capture argument is unsatisfactory when dealing with robust democracies with low levels of rent seeking and a free press to expose abuses. Yet 5.8% of MPs in New

Zealand, the country with the joint lowest perceived level of corruption in the world

(Transparency International 2016) are family followers. A more benign explanation is simply that family following occurs in politics for the same reason it does in other professions; family socialisation. Socialisation through the family is central to developing an interest in politics. Studies across different countries have found that the single most important

19 indicator of a person's likelihood of joining a political party is the previous or existing membership in that party of a close relative (Prewitt et al. 1967; Cross and Young 2008;

Recchi 1999; Van Liefferinge and Steyvers 2007). For those who go on to run for office, family socialisation is often the major factor in their choosing such a career (Clubok et al.

1969; Kurtz 1995; Cross and Young 2008; Van Liefferinge and Steyvers 2007). Indeed the child of a person who has stood for public office is significantly more likely than their peers to even consider a political career (Lawless 2012). Additionally there is a very strong positive correlation between the length of time a person spends in public office and the likelihood of being succeeded in that office by a family member (Rossi 2009; Dal Bó et al.

2009; Smith 2012).

At first glance this might appear to suggest that term limits could be an effective tool in reducing the rate of family following, but when this was tried in the Phillippines, it had the opposite effect, making political families even more powerful. The Marcos family, for example, has been able to retain a number of senior positions for decades by rotating family members through these national and state offices to circumvent the rules, an option unavailable to non-dynastic politicians (Queberin 2012).

Career following by family members is hardly unique to politics, though it does appear to be unusually high compared to other professions (Dal Bó et al. 2009: 137-138; Kurtz 1995: 448;

Laband and Lentz 1985: 413; Stephens-Davidowitz 2015). This may be related to the concept of an incumbency advantage – the clear favouritism shown by voters towards candidates

20 who are already in office. It is widely established that incumbents in democratic elections are more likely to be successful at the box (Gelman and King 1990; Levitt and Wolfram

1997). There are generally thought to be three elements to the incumbency advantage: specific office benefits that include both secretarial and other services and the higher public profile bestowed by holding office, the ability to scare off potential competitors and finally the fact that incumbents, by virtue of having been elected before and having had the benefit of on-the-job, experience are generally higher quality candidates than challengers

(Levitt and Wolfram 1997). Smith (2012) described family following as “an inherited incumbency advantage”. Feinstein’s (2010) findings support this and put the electoral benefit to the family following in the US at 4% points or roughly half the advantage traditionally conveyed by incumbency. Smith and Martin (2017) found that family followers in Ireland enjoyed a more than 35% greater chance of being elected than non-followers, but

Fiva and Smith (2016) found that though an incumbency advantage still exists in the more party centred system used in Norway, it does not appear to be heritable under that system.

Of the three components of the incumbency advantage, a part of the first – name recognition - clearly applies to family followers, but only in cases where the follower uses the same surname as their predecessor. Rossi’s (2009) study in Argentina found that family followers using the same surname were substantially more likely to be elected than those using a different name, a factor that traditionally gave a pronounced advantage to male family followers. The benefits of being in office do not strictly apply to family followers, however evidence from the US Congress suggests that politicians who hope to be succeeded by a close relative continue to engage in constituency work and other efforts associated

21 with personal vote seeking even after they have decided to retire on the assumption that the electoral reward for such work will be conferred upon the office holder’s relative

(Laband and Lentz 1985). In this, US politicians and voters are displaying a pattern of behaviour first identified in the far off days of the Roman , where Senators and their clients passed their mutually supportive relationship on from father to son (Holland

2004; Machiavelli 1503). The second explanation, the deterrent effect of incumbency does apply to family followers at least in the United States. It costs more to run against a family follower than a non-follower and the chances of defeating them are much lower (Laband and Lentz 1985; Clubok et al. 1969; Lott 1986; Lott 1989). Lott (1986; 1989) theorised that the inherited brand name advantage that family followers enjoy amounts to a sunk investment. Any challenger to a family follower must therefore spend significantly more in order to match the benefit the family follower accrues from their relative’s prior spending and activity. Curiously there does not appear be any deterrent effect in Japan (Smith 2012) despite, or perhaps because of, the much higher rate of family following prevalent there. As any Japanese citizen setting out on a political career will expect to encounter family followers wherever they choose to stand for election, there is less point in seeking out constituencies devoid of dynasties.

The third element of the incumbency advantage, quality, does not obviously apply. Family followers have not proven themselves electorally and should be in the same position in this regard as non-followers with one important caveat. Many family followers will have previously been involved in supporting the political activities of their predecessor. In

Ireland, Mary O’Rourke, a former cabinet minister and a member of a political family which

22 has produced five TDs and three Cabinet Ministers, has written about serving since childhood as a political secretary to her TD father and of how the knowledge, experience and contacts gained from that work helped her own career (O’Rourke 2012). It would seem clear then that the concept of family following as a form of inherited incumbency advantage has merit. The family follower is the beneficiary of an inter-generational wealth transfer of the goodwill acquired by their predecessor, something that is in form and nature similar to that which occurs when a farm or a shop is passed on from parent to child (Laband and

Lentz 1985).

2.2 Variations in the rate of family following between democratic states:

There are four hypotheses in the literature that seek to address the issue of variations in the rates of family following between countries. These are: that it is a cultural phenomenon associated with underdevelopment (Putnam 1976); that it is a function of the degree of openness of the electoral system (Kato S 2009; Van Liefferinge and Steyvers 2007; Dal Bó et al. 2009; Smith 2012; Van Coppenolle 2014); that there is a relationship between the rate of family following and the degree of local control of the nominating procedure (Gallagher

1988; Smith 2012; Doyle et al. 2015); or, extrapolating from Doyle et al.’s (2015) argument that family following is more prevalent among socially conservative parties, that it is a function of the prevalence of such socially conservative parties in the Parliament, though it should be noted that Doyle et al. did not draw that inference. To these four hypotheses might be added a fifth; while state capture is generally characterised as a binary variable – either the institutions of the state have been subverted to serve a powerful elite or they have not – rent seeking is a continuum, higher in some countries than in others, but

23 prevalent to some extent in all (Carson 2007). This ranges from the extreme of state capture at one end through moderately high levels of nepotistic rent seeking in functioning, but partially corrupted states such as Italy (Gagliarducci and Manacorda 2015) to much lower but still measurable forms of rent seeking by some politicians and their families even in more accountable and transparent democracies such as Sweden (Folke et al. 2015). It may therefore be possible that, if family following is linked to rent seeking, it may be higher in states where rent seeking is more prevalent and lower where it is not.

Putnam (1976) saw family following as a cultural phenomenon, a feature of what he terms

“semi-traditional societies” among which he lists “Morocco, Sri Lanka, Lebanon, Ireland,

Greece and Republican ”. Putnam did not define what he meant by a semi-traditional society or explain why family following should be higher in such states. It has been surmised, however that the very high rate of family following in the Philippines could be linked to underdevelopment (Riedinger 1995) which leads both to a lack of alternative high status careers for the relatives of the political elite, causing more of them to enter politics, while simultaneously creating barriers such as lack of education or financial resources that prevent much of the population from standing for election. A number of studies show high rates of family following in other developing countries including India (Besley 2005;

Chhibber 2011), Columbia (Acemoglu and Robinson 2006), Kazakhstan (Schatz 2004),

Mexico (Camp 1982) and Argentina (Rossi 2009) suggesting that a link does exist between low levels of development in a state and a higher rate of family following. Chhibber (2011) argues that this is in part due to the underdevelopment of the itself, with weak, under-funded political parties unable to support candidates, who must instead fall

24 back on family networks and financial support. Problematically for the theory the rate of family following in fully developed Japan at 30% (Kato S 2009; Smith 2012) is higher than in many developing countries, though rural regions of Japan do show higher rates of family following than more urban ones (Smith 2012) suggesting that rates of relative under- development within a country may be a factor.

A corollary of Putnam’s (1976) argument, that family following declines as countries develop does receive support, to a point, from the experience of the , the United

States and the Netherlands. Some 56% of all MPs serving in the UK House of Commons between 1734 and the Great Reform Act of 1832 were family followers. By 1945 this had dropped to 10% (Wasson 1991: 647). The figures from the US follow a similar pattern.

Between 1789 and 1858 11% of members of Congress were members of political families, declining to 7% in 1966 (Dal Bó et al. 2009: 119). A similar pattern is found among Dutch cabinet ministers. 61% of all Dutch ministers in the years from 1848 to 1888 came from political families, declining to 54% from 1888 – 1918, 36% in the inter-war years and 20% after 1967 (Seeker 1991: 102 – 105). These declines however ceased some decades ago.

Family following in the United States did not continue to decline after 1960 but has proved remarkably resilient (Laband and Lentz 1985; Dal Bó et al. 2009). Canada (Godwin 2013) and Australia (Lumb 2012) show a similar bottoming out of the decline in family following in the last five decades. In Ireland the rate of family following began to increase in the late

1960’s (Farrell 1970) just as the country began to modernise, while in Belgium it began to increase in the first decade of this century when the electoral system was changed in a

25 manner that increased the rate of intra-party competition (Van Liefferinge and Steyvers

2007).

Though advancing a different argument to Putnam, Smith (2012) also made the curious claim that the level of political dynasties in most democracies tends to start high and then decline. He lists two exceptions to this trend, Japan, until the 1992 reform of its electoral system, and Ireland until 2011, where he shows that the proportion of political dynasties has increased steadily. This appears to be a fundamental misunderstanding of the process of dynasty formation. By definition a dynasty cannot come into existence until a second generation replaces the first in parliament. This takes time. In Ireland, the rate of family following rose steadily until the 1969 election, some 47 years after the foundation of the state (Farrell 1970). It remained at this level with very little variation until 2011 (see chapter

5). In Japan the growth of dynasties again rose steadly for several decades, levelling out in

1993, 46 years after the Japanese Parliament was re-constituted after the Second World

War and remaining at this level for many years (Kato S 2009; Smith 2012). The decline in the proportion of dynasties in the US, the UK and Canada in the early to mid 20th century are almost certainly the result of other factors; the extension of the franchise, increasing access to education and the modernisation and development of those societies. For the last fifty years however the rate of family following in these countries has been static, albeit at a much lower level than in either Ireland or Japan.

Smith (2012) did find sudden sharp declines in the number of family followers on just three occasions; in Ireland following the 2011 election, and in Iceland and Japan after their 2009

26 elections. All three elections occurred in the aftermath of 2008 global financial crisis and saw significant political upheaval. In Ireland the long dominant Fianna Fáil party, which had come first in every election since 1932, was beaten into third place (Collins 2011). In Iceland the former ruling Independence party lost one third of its seats (Statistics Iceland 2010) while in Japan the ruling Liberal Democratic Party suffered the worst defeat in its history

(Smith 2012). In all three cases the decline in the proportion of family followers in parliament was linked directly to the declines of these political parties. In the case of

Ireland the number of family followers among the former opposition and Labour parties actually rose in 2011. Given that dynasties take time to form it is to be expected that a sudden shock to a country’s party system such as the rise of Syriza in Greece, En

Marche in France, the 5 Star Movement in Italy, the Scottish National Party in the UK or Sinn

Féin in Ireland might lead to a sharp, but probably temporary decline in the rate of family following in that country, but such declines are reflections of party system change not of advancing development.

Party system change does not always lead to a decline in family following. Smith (2012) found that in Italy, following the Tangentopoli scandal the rate of family following increased as a new political party, Forza Italia, recruited many of its candidates for office from among established political dynasties in the Christian Democratic party. As we will see in Chapter 5 something similar happened though on a smaller scale in Ireland with the foundation of the

Progressive Democrats in 1985 who drew many of their candidates from existing Fianna Fáil and Fine Gael dynasties.

27

The second explanation in the literature for variations in the rate of family following between countries is that there is a causal relationship between the degree of openness in the electoral system and the rate of family following. Electoral system openness in these studies is defined as the degree to which the electoral system allows individual voters to express their choice between candidates and ranges from Closed-List PR, where the voter cannot vote for an individual, to PR-STV where they can rank every candidate from every party and independents in whatever order they wish. There are two ways in which greater electoral system openness might lead to higher levels of family following. Firstly, there is evidence to suggest that voters prefer family followers to non-followers (Feinstein 2010;

Van Coppenolle 2014). It should not be surprising therefore that voters in more open systems with more latitude to choose the candidates of their choice, should opt for the relatives of former office holders.

Another way in which greater electoral system openness is seen to lead to a higher rate of family following, lies in the behaviour of politicians. An inherent assumption in this study is that that politicians are motivated by a desire to seek re-election and that this motivation affects their behaviour. This assumption is at the core of rational choice theories of political behaviour and is strongly supported in the literature (Epstein 1967 19-20; Bullock 1976;

Goodin 1982; Anagnoson 1983; Laband and Lentz 1985; Carey and Shugart 1995; Biglaiser and Mezzetti 1997; Altman and Chasquetti 2005) with many seeing the desire for re-election as among the very strongest factors affecting politicians’ behaviour (Amacher and Boyes

1978; Kalt and Zupan 1984; Weaver 1986). A counter-argument is put by Lott (1987) to the effect that ideology is a stronger guide to behaviour than the need to stand for re-election, but even this argument accepts that the need to seek re-election does affect behaviour.

28

Given this assumption, it has long been accepted that electoral systems affect the behaviour of politicians (Rae 1967; Patzelt 1997) and the idea that systems that allow for personalised voting encourage politicians to develop a personal vote as a means of assisting their own re- election is also widely supported (Sartori 1976; Fenno 1977; Anagnoson 1983; Cain et al.

1984; Carey and Shugart 1995; Norris 1997).

Personal vote seeking is done either through fiscal particularism, or where this is not possible, increased constituency work (Carey and Shugart 1995). Evidence from Scotland and Germany shows that politicians elected under a list system do less constituency work than those members of the same parliament who are directly elected by voters (Bradbury and Russell 2005; Lancanter and Patterson 1990). Israeli MKs freely assert that the reason colleagues do so little constituency casework is because there is little perceived electoral reward (Uslaner 1985), while British MPs and members of the US Congress in marginal constituencies engage in more constituency work than those in safe seats (Cain et al. 1984;

Ashworth and de Mesquita 2006).

The link between personalised voting and the seeking of a personal vote is thus clear. The argument that the seeking of a personal vote is accentuated in systems that allow for intra- party competition for votes from the general public – Open-list PR, PR-STV, SMP with open primaries – is both a logical assumption and is confirmed in the literature (Satori 1976;

Carey and Shugart 1995; Farrell and McAllister 2006; Van Liefferinge and Steyvers 2007; JCC

2010; Van Coppenolle 2014).

29

The results of a survey of Irish TDs carried out on behalf of the Joint Committee on the Constitution gives further support to the idea that intra-party competition may be the key explanatory variable. According to its findings, TDs who face intra-party competition engage in more constituency work, in other words more personal vote seeking, than those that do not. Furthermore the greater the level of intra party competition they face the more time they divert from other duties to constituency service. TDs who were the sole standard bearers for their party devoted 41% of their time to constituency work (JCC

2010), a rate above the 29.6% recorded in New Zealand (Anagnoson 1983) or the 30% in

Germany (Patzelt 1997) but below the 47% claimed by junior UK MPs (Wood and Young

1997). Irish TDs with one running mate spent 52% of their time on constituency work, those with two running mates devoted 62% of their time and those with three or more running mates, an admittedly small sample, spent fully 67% of their total time servicing their constituents (JCC 2010). As we have seen above, electoral competition increases the level of constituency service in elections under SMP also (Cain et al. 1984; Ashworth and de

Mesquita 2006), but not to anything like these levels.

That there is also a link between personal vote seeking and family following has been posited by researchers in Japan (Kato S 2009), Belgium (Van Liefferinge and Steyvers 2007;

Van Coppenolle 2014) and the United States (Dal Bó et al. 2009), and in the two comparative cross-national studies to date (Smith 2012; Doyle et al. 2015).

The extent to which various electoral systems incentivise the seeking of personal votes is still a matter of debate, though the area of disagreement is narrowing. Carey and Shugart

30

(1995) chose Closed-List PR and SMP with party endorsement, which they consider to be a form of Closed-list PR with a single candidate on each list, as the systems that least incentivises personal vote seeking. They held that SMP with open primaries, as used in the

United States and the Single Non-Transferable Vote (SNTV) then used in Japan are the systems in operation in developed democratic countries that most strongly incentivise personal vote seeking while considering the electoral system in the Philippines to be more open still (Carey and Shugart 1995). According to their analysis PR-STV ranked at the midpoint of the scale and was considered less open than some forms of Open-List PR (Carey and Shugart 1995). Shugart (2001) has since updated this analysis, simplifying and changing the ranking. He has kept Closed-List as the most closed, followed in order of increasing electoral system openness by SMP, PR-STV and Open-List PR while the Philippine electoral system is still considered the most open. Thus PR-STV has moved from being considered a less open system than the form of SMP which operates in the United States to being considered a more open system, but still less open than Open-List PR. Farrell and

McAllister, who strongly disagreed with the original ranking of PR-STV - describing as

“inconceivable” that PR-STV promotes less “personal vote chasing” than SMP (Farrell and

McAllister 2006) - continue to challenge Shugart’s claim that PR-STV is less open than Open-

List PR. In Farrell and McAllister’s (2006) ranking of electoral systems in order of electoral system openness, PR-STV is identified as the most candidate-centred of all electoral systems.

Colomer (2011) took a different approach. He assigned electoral systems into three broad categories: closed, semi-open and open. The inclusion of Closed-List PR and SMP in the

31 closed category is in line with the other studies quoted, but Colomer differs from all other studies in placing Open-List PR in the semi-open category along with SMP with open primaries, MMP and the Single Member Majoritarian system (SMM) used in France. He reserves the fully open category for just two systems: the Alternative Vote (AV) and PR-STV.

Thus it is generally accepted that greater electoral system openness leads to greater personal vote seeking and that the most open of all electoral systems are Open-List PR and

PR-STV, even if some debate remains about which comes first. It should thus follow that family following, which as we have seen is also positively correlated to personal vote seeking, should be higher under PR-STV and Open-List PR than under Closed-List PR and

SMP. It is important to note that this study differs from those quoted above in that it focuses on intra-party competition, a candidate-centred factor and not openness, which is a voter-centred factor. AV, for example, may be an open system from a voters point of view, but it is one almost completely devoid of intra-party competition.

Smith (2012) ranked electoral systems according to the degree to which they are

“candidate-centred” and rated SNTV as the most candidate-centred system followed by PR-

STV, Open-List PR, SMP and finally Closed-List PR as the least candidate-centred. Smith found a strong correlation between the degree to which an electoral system is candidate- centred and the rate of family following across six of the eight democracies he tested:

Japan, Belgium, Ireland, Norway, Canada and Italy. The two other countries studied,

Iceland and Israel, use a system of Closed-List-PR and showed a higher than expected rate of

32 family following. In Iceland’s case this was treated as an anomaly arising from its small population (Smith 2012).

Gallagher’s (1988) argument regarding local control of the nominating process was made with regard to Ireland. No empirical evidence was provided to support it, but given the importance of local politics in personal vote seeking it appears to be a reasonable assumption that the more local control that exists over the nominating procedure the greater the level of family following.

Smith (2012) examined this question further. He developed a model of legacy recruitment based on the supply and demand of legacy candidates. He argued that the greater the degree to which an electoral system was candidate-centred, the greater the potential demand for family followers from the voting public and that the greater the degree of de- centralisation of the candidate selection process the greater the supply of candidates from powerful local dynasties who could control the process. Equal importance was given to both factors in his theory (Smith 2012). Both greater candidate-centredness in the electoral system and greater decentralisation in the nominating process were predicted to lead to higher levels of family following. The degree to which an electoral system is candidate- centred is a national factor and varies between countries. The degree of local control of the nominating process is a party factor and may vary within countries.

Among the democracies looked at by Smith was Iceland and this proved problematic.

Iceland as stated above uses Closed-List PR, the least candidate-centred of all electoral

33 systems. Almost all parties in Iceland were also found to have fully de-centralised systems of candidate selection. The one exception, the Pirate Party, was still mostly de-centralised

(Smith 2012). These variables, according to Smith’s hypotheses, should have had opposing influences. The closed electoral system should have deterred family following, but the local control of candidate selection process should have encouraged it. Iceland in fact was found to have the highest rate of family following of all countries studied, in some elections hitting

50% of members of the Althing.

The rate of family following in Iceland is so high that it skewed the overall results of the study. With Iceland excluded, there is strong support for Smith’s first hypothesis – namely that “legacy candidates will be more common in candidate-centred electoral contests”

(Smith 2012). With Iceland included the support for the case is very much weaker.

The opposite is true for Smith’s second hypothesis that family following is more common in political parties with de-centralised systems of candidate selection. Smith (2012) looked at the nominating procedures used by forty nine different parties across eight democracies over a period of seventy years but did not find any significant correlation between the degree of decentralisation of the candidate selection process and rate of family following, except in Iceland which he concedes is an anomalous case. In addition to its very small population and the concentration of the majority of that population in a single urban centre,

Reykjavik, Iceland is also a remote island which has experienced little in the way of immigration. The level of inter-relationships on the island is such that it has even given rise

34 to a mobile phone app designed to prevent people inadvertently forming sexual relationships with close blood relatives (Sykes 2013). Put simply the chance of an elected being related to someone who has previously held office is significantly increased in these circumstances. Excluding Iceland, we can say that Smith (2012) found considerable evidence that the degree to which an electoral system is candidate-centred has a significant bearing on the rate of family following, but no evidence that local control of the nominating process has any measurable effect, though he does note that the introduction of a less de- centralised nomination process by the two main parties in Japan post 2005 is correlated with a decline in family following there.

Doyle et al. (2015) also rate the local control of the nominating procedure as being a variable of approximately equal value with personalised vote seeking. In their study they use a simplified method whereby all electoral systems in which a candidate’s name appears on the ballot paper – all systems other than Closed-List PR - are treated as being equally candidate-centred. This approach is problematic in that it represents not so much a departure from the existing literature as a complete breach with it (Carey and Shugart 1995;

Shugart 2001; Farrell and McAllister 2006; Colomer 2011, Smith 2012). No justification is given for adopting such a singular approach which conflates every other electoral system in their study, running the gamut from SMP to PR-STV, as being indistinguishable.

When seeking to measure the degree of local control of the nomination procedure, Doyle et al. (2015) use a proxy measure, the degree to which the party seeks support by focusing on the personality of its leader rather than on specific . Their position is that this allows them to account for the degree of personalism/institutionalism in a party. In essence they

35 argue that the more institutionalised a party is, the more rule bound, and the less likely it is to allow local control of its nominating procedures. It is not clear that this should necessarily follow, nor indeed that a centralised, highly controlled political party would be more resistant to being controlled by vested interests and used to the benefit of certain families. It is at least arguable that a de-centralised, more openly democratic system of candidate selection might be less open to such manipulation.

Applying these two variables to the eight countries in their study Doyle et al. (2015) came up with four categories of electoral system:

i. Party Based and Institutionalised - Netherlands

ii. Party Based and Personalised – Argentina, Israel and Japanese list MPs

iii. Candidate Based and Institutionalised – UK and Ireland

iv. Candidate Based and Personalised – USA and Japanese Constituency MPs

It should be noted that Doyle et al.’s (2015) definition of a political dynasty is a narrow one, encompassing only parents and children. As such their overall figures should be lower that those found in other studies. Doyle et al. (2015) predicted that the lowest rate of family following should be found in the Netherlands and the highest rates in United States and among constituency MPs in Japan, with the other countries in between. The results appear to match the prediction reasonably well. Family Following in the Netherlands was recorded at 0.01%. List MPs in Japan showed a family following rate of 4.3% compared to 12.8% for the Constituency counterparts. US family following however at 5.5% came in at a lower rate that the 8.5% found in Ireland, the 7.26% in Argentina or the 8% in Israel.

36

Some of Doyle et al. ’s (2015) case studies are problematic. Israel, which was also used by

Smith (2012), while in many ways a vibrant democracy, is one which has existed in a state of armed conflict with its neighbours and a section of its own population for much of its recent history. These extreme conditions are likely to be reflected in its politics. It is also worth noting that Israel, which at the time of Doyle et al.’s (2015) study stood at joint 37th place in the Transparency International Corruption Perceptions Index, had one of the worst corruption scores among all developed democratic states and one that is considerably poorer than the other states looked at by Smith with the sole exception of Italy

(Transparency 2014). This raises at least the possibility that rent seeking rather than the electoral system or degree of personalisation of the political campaigning, may be a factor in the rate of family following.

What is true of Israel, is even more so of Argentina. At 107th place in the Corruption

Perceptions Index, Argentina is one of the world’s most corrupt states (Transparency 2014).

This together with it’s underdevelopment and history of military rule make it an even more unreliable comparator.

Three of Doyle et al.’s (2015) remaining findings differ from those either in Smith (2012) or in this study. As will be seen in Chapter 4, a legislator level analysis of the lower house of parliament in the Netherlands shows a current rate of family following of 1.34% using the parent-child only model advocated by Doyle et al.. While this is certainly still lower that the

2.46% parent-child rate found in the UK or the 2.07% in Canada, it is considerably higher than the 0.01% indicated above. The figure for Ireland is also higher. While the collapse in the vote for Fianna Fáil in 2011 saw the rate of family following in the Dáil fall to its lowest

37 level in 50 years, it was not as low as reported by Doyle et al. (2015). An analysis of the biographies in Nealon’s Guide to the 31st Dáil (Collins 2011) gives a notably higher parent- child family following rate of 10.2% The 2016 election has seen a dynastic comeback and the rate of parent-child family following is now 12.02%, still low in historic terms for Ireland, but significantly higher than the 3.91% found in the current US Congress. Once these figures are taken into account, only the results for Japan match the expected outcome, but here the difference is entirely explicable by variations in the electoral system, and the possible impact of adult adoptions, and not by local control of the nominating process. Doyle et al.’s thesis is thus not supported by the evidence and when Smith’s (2012) Icelandic results are controlled for, the overall argument that local control of the nominating process is a significant factor in determining the rate of family following is simply not supported.

To recap, the literature outlines five possible causal explanations for differing rates of family following between countries:

 That it is a symptom of underdevelopment

 That it results from higher levels of corruption or state capture

 That it is a product of a conservative mindset and therefore associated with right of

centre parties

 That it is a function of the degree of local control of the candidate selection process

 That it is a product of the degree to which the electoral system is candidate-centred

The first explanation has effectively been disproved, not just by the resilience in the rate of family following in the United States (Dal Bó et al. 2009), Canada (Godwin 2013) and

38

Australia (Lumb 2012) since the 1960s, but by its actual growth in recent years in developed democracies such as Belgium (Van Liefferinge and Steyvers 2007) and Norway (Smith 2012).

Gallagher’s argument about local control of the candidate selection process is difficult to test and where it has been tested the results have not supported the hypothesis (Smith

2012). In an interview conducted by this researcher with a senior official responsible for candidate selection in Ireland’s largest political party, it was made clear that the process, even when theoretically de-centralised, is in fact micro-managed to quite an extraordinary degree. Party headquarters controls the membership list and can usually predict with a high degree of accuracy who will emerge from the selection process. If they are unhappy with the potential outcome, they can intervene to nudge the decision in the direction they wish. Among the tools available to party headquarters are:

 restricting the number of candidates a local convention may select,

 specifying the gender of the candidate(s) to be selected,

 specifying the geographical area within the constituency that they must come from,

 adding additional candidates after the event,

 and in certain circumstances, removing candidates who have been validly selected.

Given the absence of reliable, comparable data on how the decisions made in “smoke filled rooms” are really arrived at in even one country, seeking to validly compare such processes cross-nationally is highly problematic. This study will therefore not attempt to do so, but will instead argue that variations in the electoral system are, in their own right, a key explantory variable in accounting for the observed cross-national variations in the rate of family following in the parliaments of mature, developed, democratic states.

39

As mentioned, this study controls for high levels of corruption and state capture by focusing solely on mature, developed, democratic states. While no evidence has been advanced in the literature that the lower levels of rent seeking corruption that can still prevail in developed states are a causal factor in family following, this possibility will be tested in this study as an alternative hypothesis as outlined below.

Similarly no evidence has been advanced in support of Doyle at al’s (2015) notion that family following is linked to Social Conservatism, but is a testable proposition and will also be tested as an alternative hypothesis by contrasting the rate of family following in socially conservative and non-socially conservative parties in the countries under study.

We are therefore left with the relative degree to which the electoral system is candidate- centred as the only explanatory cause of cross-national differences in the rate of family following in the literature for which empirical evidence has been found (Smith 2012). The theory advanced by Smith however posits that local control of the nomination process is an equal factor. In summary Smith proposed that candidate-centred electoral systems increase the demand from voters for dynastic candidates, while local contol of the nominating process increased the supply of such candidates (2012), both elements were held to work together. The second half of the theory however, as we have seen, was not supported.

Thus no fully functioning theory of why candidate-centred electoral systems lead to higher rates of family following has yet been propounded. It is that gap that this study intends to fill. It does so by focusing solely on intra-party competition as the causal factor and by

40 proposing a mechanism of political inheritance - namely the transfer of a personal political machine from one family member to the next – that is a key explanatory variable for the differences between the observed rates of family following between countries. This study proposes a new ranking of electoral systems according to the degree to which they incentivise the creation and maintenance by a member of parliament of such a personal political machine. This study adopts a systematic approach to the choice of countries and electoral systems it looks at, choosing just one representative of each major type of electoral system identified and concentrating solely on mature, developed democratic states with low levels of perceived corruption, excluding micro-states. These case studies are thus more directly comparable than those used in previous studies and better suited to testing the empirical arguments made. Finally this study analyses its findings on separate national, sub-national and cross-national bases. In doing so it attempts to establish not just the cause, but also some of the consequences of higher rates of family following.

2.3 A Theory of Dynasty Formation:

This study looks at family following in mature, developed, democratic states. Family following in non-democratic states has existed from time immemorial and may be simply considered a form of or even if, as in or Syria, the monarchy outwardly claims to be democratic. Democracy however sits at the opposite end of the scale of political systems from the monarchical principle and the persistence of family following into political office in democratic societies is more of a puzzle. As discussed above, there is ample evidence linking under-development and family following. That developed states could show similar levels of family following without the same evidence of state capture suggests that there is something else at play. The theory being advanced here

41 is that the something else is the inheritance of personal political machines by family members.

When two candidates from the same party run on the same ticket in the same constituency under a system that allows for intra-party competition they are faced with a dilemma. It is in each candidate’s self-interest to attract more, and higher preference, votes than their running mate, but neither can do so on a policy basis as they share the same platform. Their appeal must therefore be made on a personal basis, usually on the basis of constituency service. This requires the creation and maintenance of a personal political machine; a network of clients, canvassers, fundraisers, election and constituency workers whose primary loyalty is to the candidate and not to the party, along with delegate votes at selection conventions, contacts in party headquarters and name recognition with the public.

This is a very different situation than that which operates under systems that do not allow for intra-party competition where electoral machines can be controlled by the party and put at the service of whichever candidate the party selects.

Candidates with running mates have no such luxury. The existence of such personal political machines is acknowledged in the literature (Kato S 2009; Chhibber 2011; Smith 2012). Not only are such machines heritable; inherent in their functioning is the idea of protecting a political territory from fellow party members. Among the best ways for a politician to do so is to have a clearly identified successor. Where that successor is a family member, the chances of a hostile challenge to the incumbent are reduced. In short intra-party competition increases both the opportunity for family following among politicians and the benefits to politicians of pursuing such a strategy.

42

Whether this is a demand or a supply factor is debatable. It would appear to have elements of both. The ability to inherit a personal political machine certainly increases the potential supply of family followers, by lowering the barriers to participation for those related to former office holders while simultaneously raising the barrier against those who have not been so blessed by birth. If we consider a political nomination in the same manner as entry to a market in classical economic theory it could be said that family followers pay a lower cost for market access and are therefore more likely to enter the political fray. Those outside the family circle must pay a higher price for access, by constructing their own machine from scratch, and are thus less likely to enter the market. This lower price for market access increases the potential supply of dynastic candidates.

Furthermore, the very existence of an inherited machine may also increase the demand for the family follower, both from their political party who may see in an existing machine the best way of retaining a seat and of managing intra-party competition, and from the public, who may appreciate the enhanced constituency service that such a machine delivers and wish to see it continue. As such machines are both more prevalent and more valuable in electoral systems that allow for intra-party competition, it follows that the level of such competition is in itself sufficient to account for both the increased supply of and demand for family followers.

This is a refinement of the “inherited incumbency advantage” concept identified by Smith

(2012) and Feinstein (2010). Inheriting a machine is by definition inheriting a very significant incumbency advantage, but it is the degree of intra-party competition in the electoral system, not the degree of openness, that provides the incentive to create and

43 maintain such a machine in the first place. The openness of an electoral system is a measure of the the degree of voter choice it offers, while in contrast intra-party competition is a candidate-centred measure. While there will be a degree of overlap between the two, they are not co-terminus. A case in point is the Alternative Vote (AV) system used in

Australia. This is considered in the lite rature to be one of the most open electoral systems

(Farrell and McAllister 2006, Colomer 2011), however AV promotes no more intra-party competition than Single Member Plurality (SMP), one of the least open systems does and thus, according to the theory being outlined here, despite its high degree of openness, AV should not exhibit a particularly high level of family following.

It is important to note that the level of intra-party competition can vary across and within parties even within the same electoral system. In Ireland the larger parties traditionally run multiple candidates in each constituency while smaller parties do not. In such circumstances the candidates of the larger parties have an enhaced incentive to create and maintain a personal political machine. Family following should thus be more prevalent in these more successful political parties. This factor will show considerable regional and even national variation. Parties that are traditionally strong in some regions, but weak in others may adapt their candidate strategy to run multiple candidates in some constituencies and only single candidates, or none at all in others. Under MMP smaller parties will not run constituency candidates in all constituencies but will seek list votes in all, while under Open-

List PR even the smallest parties often choose to advance a full slate of candidates despite having little prospect of most of them being elected.

44

As noted above, dynasties take time to form. The most common form of inheritance is parent-child. In almost all dynasties, each family member represents the same political party, though in both Ireland (see Chapter 5) and Japan (Smith 2012) and to a much lesser extent in Australia (Rodrigues and Brenton 2010) there is a history of family followers who are not nominated by their parties standing successfully as Independents against the party’s preferred candidate. It follows therefore that parties that have been electorally successful for more than a generation will be more likely to have family followers amongst their elected representatives. Political parties whose success is relatively new such as AfD in

Germany, Sinn Féin in Ireland, En Marche in France and Green parties in most countries will have relatively few family followers in their ranks as yet.

A corollary of this point, that political parties that have enjoyed long term success will have more family followers than those that have not, is that, in general, family followers will be disproportionately represented in governing parties and thus more likely to attain high office.

We are then left with a relatively straightforward thesis and a set of testable hypotheses that can be drawn from it. The thesis is that variations in the rate of family following between mature, developed, democratic states are, to a considerable extent, a function of the incentive to create and maintain a personal political machine and that therefore electoral systems which incentivise the creation and maintenance of such machines will see higher rates of family following.

45

Looking at the individual elements of an electoral system it is possible to identify a number of factors whose presence should have the effect of increasing the incentive to create and maintain a personal political machine.

1. Personalised voting is held in this study to be the most basic element in personal

vote seeking. Personalised voting is deemed to be possible whenever a voter has

the effective choice to vote for a candidate by name. This occurs in all systems other

than Closed-List PR. The incentive to create and maintain a personal political

machine should thus be at its lowest under Closed-List PR and exist, to some extent,

under all other systems. It should follow, if the theory of dynasty formation outlined

above is supported, that Closed List-PR should correlate with low levels of family

following.

2. Intra-party competition. This study holds that intra-party competition can occur at

two levels, Partial and Full. Partial intra-party competition is present in SMP (open

primaries) where candidates for the same party compete against one another for

public votes at nomination stage (Colomer 2011), and for Constituency MPs under

MMP where, as will be explained in Chapter 3, the parties electoral interests may

actually run counter to the candidates (Orsman 2000). Full intra-party competition

occurs under Open-List PR and PR-STV where parties run multiple candidates at

general election stage, voters are free to order those candidates as they wish and

that ordering has a real effect on the candidate’s prospects of election (Colomer

2011).

46

3. Preference vote seeking from other parties. This refers to the ability of voters to

give, and therefore candidates to seek, preference votes for candidates from parties

other than the party of their first choice. This occurs in the absence of intra-party

competition in Australia under AV and in the second round of elections under SMM

in France. This study holds that preference vote seeking without intra-party

competition does not increase the incentive to create or maintain a personal political

machine and will have little or no effect on the rate of family following, as such

appeals for votes are still primarily made on the basis of a party platform and not on

constituency service. In the presence of Intra-party competition however, such as

under PR-STV in Ireland, this study holds that the ability to seek preference votes

from the supporters of other parties ahead of a candidate’s own running mates

introduces a further personalised element to electoral competition. Such appeals

are often made on a localist or constituency service basis, complicating transfer

patterns and leading to a greater sense of “every candidate for themselves”. Marsh

(2007) shows that Irish voters regularly give a high 2nd and 3rd preference vote to

candidates outside the party they support with their 1st preference vote, even when

that party has additional candidates on the ballot. The higher level of personalised

competition engendered by such a system is predicted to increase the incentive to

create and maintain a personal political machine and therefore lead to higher rates

of family following.

A thesis based on the proposition that there is a direct causal relationship between the degree to which an electoral system incentivises the creation and maintenance of a

47 personal political machine and the rate of family following it that country’s lower house of parliament would result in the following hypotheses:

Hypothesis 1: Political dynasties will be more prevalent in electoral

systems that allow for personalised voting

Hypothesis 2: Electoral systems that allow for intra-party competition

will have higher rates of family following than those that do not

Hypothesis 3: Electoral systems that allow for preference vote seeking

from other parties but do not allow for intra-party competition will not

demonstrate notably higher rates of family following than other systems.

Hypothesis 4: By contrast, electoral systems that allow for intra-party

competition and preference vote seeking from other parties will show the

highest rate of all

If these hypotheses hold true, they should give rise to a number of additional observable implications:

1. That because what is being inherited is a local political machine, it follows that

instances of same-constituency family following will be higher under electoral

48

systems that incentivise the creation and maintenance of such machines than under

those that do not.

2. Similarly instances of family following under systems that incentivise the creation

and maintenance of personal political machines will tend to happen immediately

following the death, defeat or retirement of the preceding family member before

the machine has time to decay. Such direct inheritance will be less common under

electoral systems that do not so incentivise the creation and maintenance of

personal political machines.

3. That in electoral systems that allow for intra-party competition, family following will

be more prevalent among more traditionally successful parties, as such parties are

likely to evidence higher rates of intra-party competition. This should also hold true

for former members of such parties who choose to run as Independents.

4. In countries with mixed systems such as MMP, the rate of family following among

constituency MPs will be higher than among list MPs.

In addition, two other observations might be expected. Firstly if the electoral advantage enjoyed by family followers arises, as stated in the theory, from their good fortune in inheriting a functioning political machine, it should follow that this conveys an additional advantage. While their competitors must invest significant time and political capital in the early part of their careers constructing and refining their own political machines, family followers are free to concentrate instead on those activities that are more likely to lead to promotion – assisting the party in developing policy, campaigning for other candidates in elections to different tiers of government, developing relationships with the national media and groups. These and the significant contacts at party leadership level that

49 they have also likely inherited from their predecessor – what Smith and Martin (2017) identify as an “information advantage” - should mean that they achieve promotion more readily and that therefore:

5. In electoral systems that allow for intra-party competition family followers will be

disproportionately represented among high office holders such as cabinet ministers.

Finally, local political machines are by definition, local. Large elements of the sort of brokerage based constituency service they provide can also be provided by representatives of local or regional government. It is likely that those who prove themselves in the eyes of the electorate at local level are more likely to be favoured by that electorate as candidates for national office. Similarly, sitting MPs may see such local representatives as their future competition and will have an incentive to control local and regional seats in their electoral districts, either by occupying them themselves, or where this is not possible, ensuring that such seats are occupied by a loyal lieutenant, often a family member. Higher rates of intra- party competition are therefore concentrated at the local level. The dual effect of MPs, and their electorates, using local and regional government as a training ground for MPs and of political dynasties protecting their base by inserting family members into local office gives rise to our final additional observable implication, namely that:

6. The proportion of all MPs, both family followers and non-followers, with prior

experience in elected office at local or regional level will increase as the the degree

of intra-party competition increases:

50

There are a number of alternative explanations which give rise to further testable hypotheses:

Alternative Hypothesis 1: That family following, even in mature

developed, democratic states is linked to rent seeking behaviour and that

therefore states with higher perceived rates of corruption will also

demonstrate higher levels of family following.

Alternative Hypothesis 2: That family following, even in mature

developed, democratic states is linked to relative underdevelopment.

Alternative Hypothesis 3: That family following, even within mature

developed, democratic states is linked to underdevelopment and thus

will be more prevalent in rural areas.

Alternative Hypothesis 4: That the rate of family following is simply a

function of the superior quality of candidates from political families, in

particular their higher levels of education.

51

Alternative Hypothesis 5: That the rate of family following is linked to

traditionalism and will therefore be higher among socially conservative

parties regardless of the electoral system.

Alternative Hypothesis 6: That the rate of family following is a function of

the degree of electoral system openness and not of intra-party

competition.

The Theory of Dynasty Formation advanced in this chapter differs from others in the literature in that it clearly identifies a causal mechanism, separate from simple voter preference, for the higher rates of family observable in certain countries. If supported, this will not only increase understanding of how political dynasties are formed and persist, but also identify measures that could be taken to reduce the prevalence of such dynasties should that be so desired. Such possible measures will be looked at in greater detail in the concluding chapter.

To seek to establish which hypotheses hold true, they will be tested against the results of recent parliamentary elections in seven mature, developed democracies, against a recent provincial elections in each province of Australia and against all general elections held in

Ireland from 1981 – 2016.

52

Chapter 3: Case Selection and Methodology

3.1 Case Selection:

The case selection occurs at three levels: that of the state, that of the political parties within that state who hold seats in the lower house of parliament and that of the individual legislator. Lower Houses of parliament are being used because all democracies have one and in all countries that are considered democracies the lower houses are directly elected.

As outlined in Chapter 2, this study is concerned with mature developed democratic states only, excluding micro-states. To derive a complete list of such states, the following criteria have been employed:

 States that score highly on the Human Development Index

 States that are listed as Fully Free in the Freedom in the World Report

 States that have been democracies for at least 45 years

 States that have a population in excess of 1 million.

 States that have a positive (low) score on the Transparency International Corruption

Perceptions Index

All 187 countries and territories recognised by the United Nations are ranked according to the findings of the United Nations Human Development Index or UNHDI (United Nations

2016) a composite measure which includes Gross National Income per capita at purchasing power parity, life expectancy and general education levels. Any country or territory outside

53 the top 50 in the UNHDI has been excluded from this study. The choice of top 50 is arbitrary and the cut-off point could arguably be set higher or lower. The UNHDI has also been criticised for placing too much weight on GNI per capita (Wolfers 2009), but it remains the most widely used and accepted index of development.

All 187 countries and territories were also ranked according to their position on the

Freedom House Index or FHI (Freedom House 2016) which measures political and economic freedom. Any country or territory scoring a rating less than ‘Free’ on the FHI was excluded.

Like the UNHDI, the FHI has also been the subject of criticism. The FHI is accused of pursuing a neoliberal agenda that affects its findings (Giannone 2010) of being overly generous towards countries and territories such as Israel, that are allies of the United States of America and overly critical of countries and territories such as Russia, that are not (Bollen

1986; Steiner 2012) and for being too state centric and thus failing to pick up on emerging democratic trends in countries and territories such as Burma (Brooten 2013). Nonetheless, no systematic bias has been found in the FHI (Bollen 1986) and as with the UNHDI, the FHI remains the most widely used index of political and economic freedom. It should also be noted that its findings reinforce those of the UNHDI in that no country or territory outside the top 50 states on the UNHDI scored a ranking of ‘Free’ on the FHI. Of those within the top 50 on the UNHDI fully 42 ranked as ‘Free’. Of the top 35 countries and territories on the

UNHDI, only one – – is not ranked as ‘Free’, and even Singapore achieves a

‘Partially Free’ rating. Among the top 50 countries and territories only six are rated as ‘Not

Free’; resource rich Brunei, Qatar, Bahrain and the United Arab Emirates, and Hong Kong

54 and Belarus. Apart from Singapore, the only other country or territory in the top 50 rated as

‘Partially Free’ is the Seychelles.

All 187 countries and territories were also measured against the age of their democracy.

Those without democratically elected were excluded as were those in which democratic elections have been held for less than 45 years. An earlier cut-off could be adopted. Rossi has found that dynastic tendencies became clear in Argentina after just 20 years. The reason for opting for 45 years was to allow time for the majority of the first generation of parliamentarians in a democracy to retire and be replaced, thus allowing for patterns of family following to emerge. A second reason is to allow time for the institutions of the state to become established and memories of a non-democratic past to recede. As seen in Chapter 2, the number of family followers in Ireland only overtook veterans of the

War of Independence in the Dáil at the 1969 election (Farrell 1970), 47 years after the establishment of the state, while in Japan dynasties reached their peak, or rather plateau,

46 years after the establishment of the modern Japanese state (Kato S 2009; Smith 2012).

Once again the findings reinforce those of the other measures. No country that has had an uninterrupted democracy since 1973 is outside the top 50 states on the UNHDI or scores anything other than Free on the FHI.

To avoid the difficulties encountered by Smith in analysing family following in Iceland, this study will exclude microstates, defined here as all countries with a population of less than 1 million. Once again this is something of an arbitrary figure. There is no official definition of

55 a microstate, though a cut-off of 1 million has been used by scholars (Boyce and Heir 2008) and appears to have been accepted as an effective shorthand (Dumienski 2014).

Just 19 states or territories meet these four requirements. The final filter is Transparency

International’s Corruption Perception Index. As the literature attests, high levels of corruption are widely seen as a causal explanation for high levels of family following. The assumption behind applying the Corruption Perceptions Index as a filter is that countries and territories with a low level of perceived corruption, will actually have low levels of corruption and rent-seeking behaviour. Once again the Index is not without its critics. It is based on reports of perceived corruption, emanating mostly from the business community.

Among the problems associated with the Index are that because it deals in perception, the very fact of labelling a country as corrupt, may lead that country to suffer from a higher perceived level of corruption than would otherwise be the case (Warren and Laufer 2010).

It was further argued that the Index disadvantages large countries, which tend to have a higher overall number of reported cases, over small ones who, despite having fewer cases may suffer from higher relative levels of corruption (Saisana and Saltelli 2012). A final critique is that because aid donors and investors place so much emphasis on the Corruption

Perceptions Index, that it may create a “corruption trap” whereby poor countries, who tend to suffer most from corruption, receive less aid and investment (Warren and Laufer 2010;

Andersson and Heywood 2009) and are therefore less able to combat corruption, though this is less a criticism of the accuracy of the Index, than of the use to which it is put.

56

In response to these criticisms, Transparency International changed its methodology in

2012. The new methodology has been found to be more robust and free from significant bias (Saisana and Saltelli 2012) and it remains the most widely used and trusted measure of political corruption in the world.

The choice of the cut-off point as the top 25 least corrupt countries is, like all the others, arbitrary and could be set higher or lower. Just two of our 19 countries, Italy and Israel, score outside the top 25 least corrupt nations by Transparency International Corruption

Perception Index (Transparency 2016). Israel, after a significant improvement in its performance in 2016 comes close to the cut-off, at 28th in the world, but given the other issues affecting Israel outlined in Chapter 2, it is felt safer to exclude the Knesset from this study. Italy’s problems with preceived corruption are more serious and more ingrained.

Currently ranked at 60th in the world, Italy has the same level of perceived corruption as

Cuba and scores below many developing countries, the only developed democratic state to consistently do so (Transparency 2016).

When Italy and Israel are removed we are left with a list of seventeen mature, developed, democratic states using eight broad types of electoral system. Ranked from left to right according to the openness of thee electoral systems used to elect the lower houses of their parliaments according to Farrell and McAllister (2006) these are shown in Table 1 below:

57

Table 1: Electoral systems used in Mature, Developed, Democratic states

Closed- SMP (open SMM (2 Alternative Open-List List PR SMP primary) Rounds) MMP Vote PR PR-STV United Austria Canada States France Germany Australia Belgium Ireland United Kingdom Japan Denmark New Zealand Finland Switzerland* Netherlands Norway Sweden Source: Inter-Parliamentary Union Parline Database http://www.ipu.org/parline-e/parlinesearch.asp

* the system used in Switzerland is classfied by the IPU as a “Mixed” sytem, in that voters can either vote for a party list or for an individual

candidate. Unlike MMP however, voters only have a single vote.

This study uses the Farrell and McAllister (2006) model of electoral system openness as a

starting point, but as discussed in Chapter 2, the key variable hypothesised to lead to higher

rates of family following is not openness, which is voter centred but the degree to which the

electoral system incentivises a candidate to create and maintain a personal political

machine. This factor, which is candidate-centred, is held, as outlined in the Theory of

Dynasty Formation, to have three component parts: personalised voting, intra-party

competition and preference vote seeking from other parties and will be represented by the

symbol i .

For example , the SMP (open primary) system used in the United States gives most voters

the opportunity to vote twice; once to select a party candidate and once to elect a person

to office. In states with closed primaries, those voters who choose to register as

58

Independents may only vote in the second ballot. Looked at from the voter’s point of view this system belongs where Farrell and McAllister put it; more open than the form of SMP used in the UK and Canada, but certainly less open than MMP where every voter gets two votes, and less open than the SMM where voters get the chance to vote a second time if no candidate exceeds 50%.

Looked at from a candidate point of view however, SMP (open primary) is a very different beast. A candidate seeking election under SMM, or as a constituency member under MMP will, like their UK and Canadian SMP counterparts, be their party’s sole standard bearer. If a list candidate under MMP, they will have even less incentive to create and maintain a personal political machine. SMP (open primary) by contrast allows for full-blooded if time- limited intra-party competition. The scale of many such primaries and the resources dedicated to winning them are of an order of magnitude greater that what would be expended in securing a party nomination under any of the other systems. Some 60% of eligible US voters participated in final primaries in the 2008 presidential election (Colomer

2011). Voter participation in Congressional primaries, particularly in non-presidential years is much lower, but is still far beyond that seen elsewhere. Some 18 million US citizens voted in Congressional primaries in 2014 (Schwarz 2014). Securing a nomination thus requires the creation and maintenance of a significant personal political machine. While this machine may be enhanced by the resources of the party once the nomination has been secured, few

US politicians can afford to rely solely on their party.

59

The growing tendency for gerrymandering Congressional constituencies exacerbates the level of intra-party competition. According to Colomer (2011) half of all US Congressional seats are safe for one party or the other. Indeed in 2014 some twenty eight seats in the US

House of Representatives were uncontested by the opposing party at the general election, a phenomenon found in no other country in this study. If the general election in a district is a foregone conclusion, there is less incentive towards party cohesion and a greater likelihood that even an incumbent will face a primary challenger. Twelve of the twenty eight members of Congress whose seats were so safe that they were unchallenged by the opposing party in

2014 had still to endure primary challenges, two of which were successful. According to the theory outlined in Chapter 2, intra-party competition is a more powerful causal factor in raising the level of i than either personalised voting or preference vote seeking from other parties. Accordingly in this study the SMP (open primary) system is classified as having

Partial Intra-Party Competition and is thus held to have a third highest degree of i, lower only than Open-List PR and PR-STV.

A second criticism of both the Farrell and McAllister (2006) and Colomer (2011) models is that there is no real justification for ranking the AV system used in Australia as more open than the Single Member Majoritarian (SMM) system used is France. Both systems work in a similar way. The voter is asked to pick the candidate of their choice in a single seat constituency. If their chosen candidate is not elected and no other candidate has exceeded

50% of the vote, the voter is then asked for their second preference. There is really only one difference. Under AV the second, and subsequent, preferences are cast at the same time as the first, while under SMM it is cast some time later, usually after an interval of a

60 few weeks. From the point of view of the voter, both systems are essentially the same, though under SMM the voter does have better information in the second round. From the point of view of the candidate seeking election there are two elements to be considered which have oppossing effects:

i. It is arguable that the delayed second vote under SMM allows for additional

personal vote seeking from voters who would not normally support the

candidate, something which should lead to SMM being rated as more candidate-

centred than AV

ii. Under AV all candidates have the opportunity to canvass for lower preference

votes, under SMM only the top two do. This ability for all candidates to

potentially benefit from preference vote seeking from other parties should lead

to AV elections showing higher levels of personal vote seeking overall than SMM

elections

Under both systems the incentive for a candidate to create and maintain a personal political machine separate from their party - i - is decidedly limited and no substantial difference appears to exist between them. In both cases the candidates remain the sole standard bearers for their parties. Thus, while there may be some incentive to cultivate a personal vote, there is little incentive to create and maintain a personal political machine.

61

Given the similarities between the effect on i of both SMM and AV, for the purposes of this study they are considered to be variations of the same system. Given the low level of i under both, we should further expect the level of family following under either SMM or

AV will not be significantly higher than that seen under SMP.

Farrell and McAllister (2006) list MMP as the next most open system, as the existence of two votes, one for a constituency representative and one for a national party list, provides a greater degree of voter choice. This study agrees with the ranking, but for quite different reasons. At first glance, from a candidate point of view, MMP appears to be a combination of Closed-List PR and SMP and one might reasonably expect that it would be ranked, in terms of i, between these two systems. In practice it works quite differently. While candidates on the party lists do not have any incentive to create and maintain a personal political machine, this study argues that constituency candidates under MMP have a significantly greater incentive to do so than their counterparts in SMP elections in the UK and Canada.

In New Zealand for example, the total number of seats won by a party in parliament is equal to the % vote won by that party in the list ballot providing that the party vote exceeds a 5% threshold (New Zealand Electoral Commission 2017). Any constituency seats won do not add to the total seats the party has in Parliament, but rather come at the cost of List seats.

The Electoral Commission of New Zealand provides a useful seat calculator on its website.

62

To illustrate how seats are allocated, let us imagine a simplified New Zealand election in which only the Labour and National parties run candidates. In this imagined scenario, let us assume that both parties win 50% of the list vote, but that the National Party outperforms

Labour by a wide margin in the Constituency seats, say by 5:1. What would be outcome of such a hypothetical contest?

According to the Electoral Commission, both parties would emerge with the same number of MPs. The National’s success in the constituencies would simply lead to them cannibalising their party list. In other words there is little incentive, from a party point of view, in performing well in the constituencies. It is the list vote that counts.

There is an exception in New Zealand in the case of small parties who do not reach the 5% threshold. Should such parties manage to win a constituency seat, the threshold is waived and the party gains an additional % of seats equal to their national list vote. If this results in more than 120 MPs being returned, additional seats are created. While this provides an incentive for small parties to provide the maximum support to their constituency candidates, for larger parties, it may in fact create a perverse incentive in some cases to sabotage their own standard bearers. A case in point is the 1999 General Election when on the eve of the vote, the Labour party leader openly encouraged voters to abandon her party’s candidate in a marginal constituency in favour of the (Orsman 2000).

The Greens duly won that seat and joined Labour in government. While in most cases, parties in New Zealand campaign for the highest possible vote in both list and constitiuency

63 competition, the peculiarities of their system mean that constituency candidates have a strong incentive to create and maintain a personal political machine independent of a party headquarters that could turn hostile to their interests. Add to this the tendency of voters under MMP to ticket split (Colomer 2011) and it becomes at least arguable that a strong performance in constituency seats for a major party might result in that party having fewer

MPs in Parliament.

As with the SMP open primary system used in the United States the level of intra-party competition among constituency MPs in New Zealand can be considered as Partial, rather than Full. Not all parties are affected and not all candidates may fear finding themselves on the wrong side of their party’s electoral strategy. By contrast candidates with running mates in elections held under Open-List PR or PRSTV are under no such illusions and must create and maintain a personal political machine or risk losing their seats.

Finally, in our ranking of electoral systems, it should be noted that the degree of i in

Flexible and Open-List PR varies considerably. While in Belgium candidates are elected on a local basis and voters have considerable latitude in ordering the candidates within the party list, in neighbouring Netherlands the system is far more restrictive. Although in theory MPs have sub-national geographical constituencies, in practice votes are counted and seats are allocated on a national basis. Voters may alter at most the ranking of two candidates on the list, a degree of openness that has little practical effect. These restrictions, as Colomer

(2011) noted result in a much lower level of preference voting in the Netherlands than

64 occurs in Belgium and helps explain why Doyle et al. (2015) in their study treat the

Netherlands as a form of Closed-List PR. Leenknegt and van der Schyff (2007) go further, describing the 1989 reform of the Dutch political system, which was designed to enable greater voter choice over the election of candidates within party lists, as a failure. They found that Dutch MPs loyalty is to their party, not the electorate and that the vast majority of Dutch MPs are in fact unknown to the general public (Leenknegt and van der Schyff

2007). Unhappiness with the lack of effect of the 1989 reforms led to a renewed effort to change the electoral system to MMP in 2005 but this was rejected by the Council of State. It remains very difficult in the Netherlands to get elected on the basis of personal vote seeking. In elections since 1989 the total number of Dutch MPs elected on the basis of a personal vote has never exceeded two (Renwick and Pilet 2016:242). The biographies of the members on the Parliament website in the Netherlands don’t even mention the constituencies the MPs supposedly represent, such information having little practical relevance. From the candidate point of view therefore intra-party competition for personal votes in the Netherlands is, along with Austria, the lowest among the list of countries in this study.

If our ranking of electoral systems is adjusted to take these factors into account, it now looks like this with i increasing as we move left to right

65

Table 2: Electoral systems used in Mature, Developed, Democratic states ranked by Incentive to Create and Maintain a Personal Political Machine

SMM (2 Rounds)/ Closed-List Alternative SMP (open Open- PR SMP Vote MMP primary) List PR PR-STV United Austria Canada France Germany States Norway Ireland United New Netherlands Kingdom Australia Zealand Sweden Japan Denmark Switzerland Belgium Finland i

This then leaves us with seven broadly defined electoral systems to examine. Closed –List

PR clearly allows for the least degree of personalised vote seeking. SMP as practiced in

Canada and the UK is broadly similar. Voters do vote for a candidate and therefore there is some incentive to foster a personal vote, but the lack of both intra-party competition and preference vote seeking from other parties limits the effectiveness of such a strategy. SMM and AV both allow for full preference vote seeking from other parties but not intra-party competition increasing their degree of i very modestly, if at all. MMP comes next. As outlined above, the existence of a separate party vote reduces the party’s incentive to support constituency candidates and therefore increases the constituency candidate’s need to create and maintain a personal political machine. In addition the separate vote allows for the possibility that candidates for election to constituencies might appeal for local votes from supporters of other parties who would still be free to cast their list vote for the party of their choice. The evidence from Japan, Scotland and Germany certainly indicates a

66 marked difference between the behaviour of constituency and list MPs (Bradbury and

Russell 2005; Lancanter and Patterson 1990; Smith 2012; Doyle 2015) .

SMP (open primary) shows the 3rd highest incentive to create and maintain a personal political machine, a reflection of the greater, if still limited, degree of intra-party competition under this system, a degree of intra-party competition made appreciably more severe in hyper-partisan districts.

As discussed above, there are differences between the various methods of Open-List PR, principally concerning constituency size and the manner in which preferences for individual candidates can be expressed. Despite this, as we have seen in the literature, this system is held to incentivise personalised vote seeking to a greater extent than any other except PR-

STV which, by adding preference vote seeking from other parties becomes the system most orientated to the creation and maintenance of personal political machines and therefore the system likely to yield the highest rate of family following.

Ideally all the states listed should be looked at, but information on family following in all of them is not readily available. In consequence this study will look at one state from each electoral system category, choosing on the basis of convenience and accessibility.

67

Only the United States uses SMP with open primaries and, of the qualifying countries only

Ireland uses PR-STV, meaning both countries are automatically selected. The Netherlands has been chosen over Austria as the exemplar of Closed-List PR due to the superior quality of the biographical information available on the Dutch parliament website. Canada has been chosen over the UK for the same reason. While Dodds (2011) Parliamentary

Companion provides biographical information on all UK MPs, family connections, other than parent-child, are not always included. By contrast, the Parliament of Canada (2016) website provides a searchable database documenting all the family connections that this study is looking at. It should be noted that an analysis by this researcher of the biographies of those

MPs elected to the UK House of Commons in 2010 using Dodds (2011) showed a parent- child family following rate of 3%, broadly comparable with the 2% rate of parent-child following found in the current Canadian House of Commons.

Australia has been included at the expense of France, primarily due to the existence of two recent studies providing detailed analyses of family following in the Australian Parliament

(Lumb 2012; Lumb and Goddett 2015).

New Zealand is being used as the model for MMP. As outlined, above the issue of adult adoptions complicates the data on Japan and while Switzerland is classified by the

Interparliamentary Union as a Mixed system (IPU 2016), its operation is quite different to

MMP. Germany uses the classic form of MMP but while most of Germany has been a democracy for more than forty five years, the eastern part of the country only became

68 democratic in 1990 a fact which complicates cross-national comparisons. In addition the extra focus in New Zealand on their electoral system among academics and commentators caused by recent referendum on the continued use of MMP in New Zealand has aided the choice.

Of the countries using various forms of Flexible and Open-List PR, Belgium provides the greatest incentive to cultivate a personal vote and thus the greatest incentive to create and maintain a personal political machine. Not only are Belgian voters more likely to cast a personal vote than voters in any other mature, developed democratic state using List PR, those personal votes are also more likely to determine which candidates are elected

(Renwick and Pilet 2016: 219-242). The Swedish electoral systems by contrast is almost as restrictive as that used in the Netherlands. Voters may cast a single vote on either a party or a candidate basis, though crucially seats are allocated on a county rather than a national basis making personal vote seeking a more viable candidate strategy; not much more viable though, in 2006 just six members of the Riksdag were elected as a result of candidate- centred voting (Renwick and Pilet 2016: 147).

Finland and Denmark present different problems. In Denmark the parties decide whether to use an open or restricted list and different parties in the same constituency will chose different approaches (Renwick and Pilet 2016: 239) adding an extra level of complexity for a comparative study such as this. The Finnish system appears very open but the problem here is that there is simply no agreement among those who have studied the country on the

69 extent to which candidates are elected as a result of personalised voting (Renwick and Pilet

2016: 180). Norway’s classification is a matter of debate. The system as described in the

Representation of the People Act 2002 and in Norwegian government guidelines is clearly

Open-List; elections are held and seats allocated in local constituencies and voters are empowered to alter the order of the names on the party list (Government Norway 2009;

Representation of the People Act 2002 Chapter 7,2), however in much of the literature the

Norwegian Electoral system is described and treated as a form of Closed-List PR (Smith

2012; Enli and Skogerbo 2013; Fiva and Smith 2016). Renwick and Pilet found that although the Norwegian system does allow voters considerable freedom in who they vote for it is structured in such a way that these votes have no practical effect (Renwick and Pilet 2016:

151). Norway also weights its parliament to give more representation to rural areas

(Government Norway 2009; Representation of the People Act 2002) something that could have a distorting effect on cross-national comparisons. For these reasons, Belgium has been selected as the model for Open-List PR.

The final list of countries that are included in this study are outlined in Table 3 below ranked from left to right in ascending order by the degree to which their electoral systems incentivise the creation and maintenance of a personal political machine: i . If the Theory of Dynasty Formation outlined in Chapter 2 is correct, then the results of our legislator level analysis should show an increasing rate of family following as we move from left to right across the table.

70

Table 3: Countries in Study by Electoral System

SMM(2 Closed-List Rounds)/ SMP (open Open-List SMP MMP PR-STV PR Alternative primary) PR Vote MP

New Netherlands Canada Australia United States Belgium Ireland Zealand

i

As discussed above the incentive to create and maintain a personal political machine will be

much lower for those New Zealand MPs who are elected on a party list than for those who

seek to be elected in constituencies. If New Zealand were to be treated as two systems,

then, as in the Table 4 below, List MPs in New Zealand should be expected to show a lower

rate of family following than MPs in Canada elected under SMP and Constituency MPs in

New Zealand should manifest a rate of family following similar to that of members of the US

House of Representatives.

3.2 Predictions:

Another way of thinking about the electoral systems in this study, is to rank them according

to the presence, or absence, of the four key electoral system features indentified above

namely:

71

 Personalised Voting: represented below as PV

 Preference vote seeking from other parties: represented as ç

 Partial Intra-party Competition: represented as PIPC

 Full Intra-party Competition: represented as FIPC

Again the incentive to create and maintain a personal machine and therefore the rate of family following should increase from left to right as we move across the table.

Table 4: Countries in Study by Electoral System Features

United States/New Netherlans/NZ List Canada Australia Zealand Belgium Ireland SMP (open Alternative primary)/ MMP Open- Closed-List PR SMP Vote Constituency List PR PR-STV PV + No PV PV PV + ç PV + PIPC FIPC PV+ FIPC + ç i

As can be seen in the table above. Closed-List PR and the list element of MMP will display none of these features, while PR-STV has all of them. The other electoral systems all have at least one feature and, according to the Theory of Dynasty Formation outlined in Chapter 2, both the incentive to create and maintain a personal political machine and the rate of family following should increase as we move from left to right on the table and the more significant elements, namely higher degrees of intra-party competition, come into play. This

72 prediction will be compared to the actual findings from a recent election in each of the countries in this study in Chapter 4.

3.3 Methodology:

The results of a recent election from each country will be examined in detail. The findings will be subject to analysis on three levels:

i. Individual Legislator

ii. Political Party

iii. National

3.3.1 Legislator Level Analysis

The 1,502 legislators elected in a recent election to one of the seven lower Houses of

Parliament in this study each constitute a single case. All are coded for the following 5 attributes. A binary method has been used with 1 indicating the presence of a variable and 0 its absence:

 Family following. Is the legislator a family follower: i.e. are they closely related to a

current or former member of the same parliament and were they first elected to that

parliament after the election of their close relative?

 Social conservatism: does the MP represent a political party deemed to be socially

conservative? See 3.2.2 below.

 Gender. Two interpretations of a gender gap are possible. If women are significantly

more highly represented among family followers than non-followers, it may indicate

that some of the barriers that inhibit greater female participation more generally do

73

not affect family followers to the same extent (Folke et al. 2016). If the opposite is

the case and women are under-represented among family followers to a significant

level, this may be evidence that even in developed democracies family following may

be a symptom of a traditionalist mindset.

 Whether the MP is a graduate. This is to test both Hess’ (1997) hypothesis that

family followers are better educated than non-followers and Feinstein (2010) and

Danielle and Geys (2014) contrary argument that family followers are in fact less

qualified than non-followers. For reasons of simplicity and comparability a single

measure is used; has the MP been awarded a university degree or equivalent

professional qualification or not.

 Prior experience at local or regional level. This is predicted to manifest itself in

different ways in different countries. In those with lower levels of intra-party

competition, family followers may, if they do enjoy an advantage over their peers,

skip the sub-national level and move directly into parliamentary politics. However in

systems that allow intra-party competition at the constituency level, the opposite

should be the case; local office will in many cases be occupied by an aspiring dynast

to protect their MP relative’s political base and assist their own eventual succession.

One other attribute will be recorded for all 1,502 cases:

 Age at first election. Dal Bó et al. found that family followers in the United States

tended to have longer careers than non-followers. Age of first election is a proxy

measure for this in existing parliaments where many MPs may still be in the early

stages of their careers (Dal Bó et al. 2009).

74

For legislators who are identified as family followers, additional information was

gathered to identify and code the nature of the relationship – child, sibling, spouse etc. –

and the form of inheritance – direct, delayed or extra-constituency. The data thus

provided will be subject to regression analysis and analysed both sub-nationally and on a

comparative cross-national basis. In the longitudinal study of Ireland these attributes

will be measured over time.

3.3.2 Constituencies:

All constituencies in Canada, Australia, New Zealand, the United States, Belgium and Ireland have been categorised as either:

 Urban: population density in excess of 500 persons per square kilometre

 Rural: population density of less than 150 persons per square kilometre and

not having a single urban settlement of more than 50,000 people

 Mixed: population density between 150 and 500 persons per square

kilometre or of less than 150 persons per square kilometre but containing a

single urban settlement of more than 50,000 people.

These measures are based on the new degree of urbanisation categorisation DEGURBA used by the OECD (Djikstra and Poelman 2014).

In terms of urbanism, list MPs from the Netherlands and New Zealand have been coded as having a mixed constituency. Morris (2010) suggests urbanism is a useful proxy for development. This measure will be used to establish whether there is a link between lower

75 levels of development within a developed state and family following. If true, less developed rural areas of developed states should elect family followers more often than urban areas.

3.3.3 Party Level analysis:

Two variables will be measured on a party political basis – ideology and whether the party has enjoyed significant electoral success over a long period of time.

Doyle et al. (2015) suggested that more traditionalist centre-right parties might produce more family followers. This seems to be a modification of the Putnam (1976) argument positing a link between traditional societies and family following. If so, the key question is whether a political party is associated with conservative social thought.

The European Election and Referendum Database (EED 2017) categorises all political parties in Europe by ideology. Any political party in the Netherlands, Belgium or Ireland that is categorised as conservative or Christian democratic by the database is coded as socially conservative in this study. Political parties in these countries that are categorised as socialist, social democratic or green by the database are coded as not being socially conservative. A number of liberal, regional/nationalist and special interest parties, such as

Vlaams Belang are classified as socially conservative by the EED and are coded as such.

Those liberal, regional/nationalist and special interest parties that are not classified as socially conservative by the EED are treated on a case by case basis as outlined below:

76

 Netherlands: Volkspartij voor Vrijheid en Democratie – VVD - is classified as liberal in

the EED, but coded as socially conservative in this study.

 Belgium: The New Flemish Alliance (NVA) is classified as ethnic regional, but coded

as socially conservative in this study. The Mouvement Réformateur (MR) and Open

Vlaamse Liberalen en Democraten (Open VLD) parties are both classified as liberal

and are coded as not socially conservative. The Libertair, Direct, Democratisch (LDD)

group is classified as liberal/free market and is coded as socially conservative.

 Ireland: The Sinn Féin party are classified as special issue but are coded as not

socially conservative. Independent TDs who are former members of political parties

are coded in the same manner as their former parties.

For countries outside of Europe the following designations have been adopted:

 Canada: The Conservative party is coded as socially conservative. The Liberal, New

Democratic and Bloc Quebecois parties are coded as not being socially conservative

 Australia: The Liberal, Country Liberal, National, Katter Australian and Palmer United

parties are coded as socially conservative. The Labour and Green parties are coded

as not being socially conservative

 New Zealand: The National, New Zealand First, and Association of Consumers and

Taxpayers – ACT – political parties have been coded as socially conservative. The

Labour, Green and Maori parties have been coded as not being socially conservative.

 United States: The Republican party has been coded as socially conservative. The

Democratic party has been coded as not being socially conservative.

77

The reasons for looking at parties’ electoral fortunes over time is to test additional observable implication 4 which states that family following in electoral systems that allow for intra-party competition should be more prevalent among traditionally successful parties.

For the sake of this study political parties that are coded as “traditionally successful” will be parties that have had continuous representation in the lower house of their national parliament for at least 25 years and that have been either the largest or second largest parties in that parliament during some or all of that time period. This variable will only be coded for those electoral systems that allow some degree of intra-party competition and thus excludes Netherlands, Canada and Australia.

The following parties have been classified as traditionally successful for the purposes of this study

 New Zealand: The Labour and National parties are traditionally successful. New

Zealand First, ACT, United Future, Green and Maori parties hold seats in parliament

but have either not had continuous representation for 25 years or have not been the

largest or second largest parties at some point during that time

 United States: The Republican and Democratic parties are both traditionally

successful. No other parties hold seats in the House of Representatives.

 Belgium: The Mouvement Réformateur (MR), Parti Socialiste (PS), Christen-

Democratisch en Vlaams (CDandV), Open Vlaamse Liberalen en Democraten (Open

VLD) and The New Flemish Alliance (NVA) – are classed as traditionally successful.

78

All have been in parliament for at least 25 years and all have been the largest or

second largest party in parliament at some point over that period. The Libertair,

Direct, Democratisch (LDD), Ecolo, Groen, Socialistische Partij Anders (sp.a), centre

démocrate Humaniste (cdH), Fédéralistes Démocrates Francophones (FDF) and

Vlaams Belang all have seats in parliament but are not classified as being

traditionally successful.

 Ireland: Fine Gael, Fianna Fáil and Labour are the traditionally successful parties.

Sinn Féin, the Anti Austerity Alliance/People Before Profit (since re-named

Solidarity), the Social Democrats and the Green party also hold seats in parliament,

though none has done so continuously for 25 years or been amongst the two largest

parties during that time.

3.4 Sources for countries in study:

3.4.1 Netherlands

Sources:

The website of the House of Representatives (Parliament of the Netherlands 2016) provides detailed biographical information on each MP including their date of birth, educational qualifications, previous experience and – where appropriate – their familial relations to previous members of Parliament or government. This information was cross referenced with the biographies of the MPs on their own and their parties’ websites. The information thus secured is robust and if any systematic errors exist, their effect would be to understate

79 rather than overstate the rate of family following. The general election looked at was that held in 2012.

Dutch MPs are technically elected in regional constituencies, but votes are counted and seats allocated at a national level. Curiously neither the parliament website nor the party or

MP websites make any mention of the constituencies for which the MPs are nominally elected. This re-inforces the point made by Doyle et al. (2015), Colomer (2011) and

Leenknegt and van der Schyff (2007) that although in theory, the Dutch use a form of limited Open-List PR, in practice theirs is, in effect, a closed system. Because of this lack of constituency focus, questions relating to constituencies or whether the seat inherited was in the same constituency as that held by the preceding family member have been omitted. For the question on urbanism, all Dutch constituencies have been coded as mixed. See the methodology section earlier in this chapter for further details.

It should also be noted that Netherlands requires MPs who are appointed to cabinet to vacate their seats in parliament, thus no MPs in parliament are current ministers. The Dutch also appoint many ministers from outside parliament. The effect of these factors is to reduce the number of MPs in parliament who are former ministers when compared with the other countries under study.

3.4.2 Canada

Sources:

The Parliament of Canada (2015) website provides detailed biographical information on all current and former MPs going back to the founding of the Dominion of Canada including

80 information on education and previous experience. It also includes a searchable online database dedicated to family connections among MPs. This information was cross referenced with party and personal websites. Dates of birth are not always given and in some instances the age of the MP on first election has been estimated from their biographies. The Elections Canada (2015) website provides detailed maps and demographic information for each parliamentary constituency. The general election looked at was the one held in 2015.

3.4.3 Australia

Sources:

The primary source for Australia was an updated version of Martin Lumb’s (2012) study into family following in the Federal parliament which covered the parliament elected in 2013

(Lumb and Goddett 2015). The Parliament of Australia (2013) website provides detailed demographic information on all constituencies and comprehensive biographical information on all MPs, but omits family connections. While such information can be sourced from party and MP websites, it is not definitive. For this reason the figures for

Australia relate to the 2013 Federal election rather than the more recent 2016 vote. The results are consistent with a steady rate of family following in Australia as observed by Lumb over the last number of decades (Lumb 2012).

As Australia uses both PR-STV and AV at state level, a further analysis was carried out on the results of the provincial elections held in Western Australia in 2013, Tasmania in 2014,

South Australia 2014, Victoria in 2015, Queensland in 2015 and New South Wales 2015.

This analysis compares the rate of family following only between the lower houses of the

81 various state parliaments. The information is sourced from candidate and party websites and regional news media. If there is any systematic error it will be to understate the rate of family following in each case.

3.4.4 New Zealand

Sources:

The Parliament of New Zealand (2016) website contains biographical information about each member of parliament and for constituency MPs, detailed demographic information about their constituencies. The information does not include family connections to current or former MPs. This information was sourced from party and MP websites and articles in

New Zealand media. Every effort has been made to ensure the accuracy of the data. If there are any systematic errors the effect of these would be to understate the rate of family following. The general election looked at was the one held in 2014.

3.4.5 United States

Sources:

The website of the US Congress (2016) hosts a searchable database of all members going back to the pre-Independence Continental Congress giving details of age, education, occupation and familial relationships to other members of Congress. The US Census department provides detailed maps and demographic information on all congressional districts. The information thus provided has been cross referenced with the individual and party websites of each member elected to the House of Representatives in 2014. As the US has a presidential rather than a , the question on cabinet membership has been omitted from the US portion of this study. As noted above Presidential elections

82 can have a significant effect on voter turnout. To control for this the US election included in this study is the last mid-term race in 2014.

3.4.6 Belgium

Sources:

The Parliament of Belgium (2016) website provides detailed biographical information on all current and former members of Parliament going back to 1991. This includes date of birth, education, roles in local, regional and national government and family relationships to elected politicians at local, regional and national level. This information has been cross- referenced with candidate and party websites and Belgian media reports. The websites of the ten Belgian regions give detailed demographic information on the constituencies. The general election looked at was the one held in 2014.

3.4.7 Ireland

Sources:

Nealon’s Guide (Collins 2016) is a biographical dictionary of the members of the Irish Dáil and Seanad published after every general election since 1973. The guide provides information on TDs dates of birth, education, current and previous roles and their familial relationship to current and former members of the Oireachtas. This information has been cross referenced with candidate and party websites and with the Irish media. The constituency information is sourced from the Central Statistics Office and from the website of Dr Adrian Kavanagh (Kavanagh 2016). The general election looked at was the one held in

2016.

83

Chapter 4 Results

In this chapter the results of a recent general election to the lower house of parliament in each of the seven countries in this study will be used to test the Theory of Dynasty

Formation outlined in Chapter 2.

Section 4.1 will deal with the overall headline results and compare these results, by country and by electoral system, to the predictions that arise from the Theory of Dynasty Formation.

Section 4.2 will examine the findings from each country in detail. Section 4.3 tests those findings on a comparative cross-national factor basis against the various hypotheses that underpin the Theory of Dynasty Formation. Section 4.4 tests these national and cross- national findings against the set of additional observable implications that arise from the theory as outlined in Chapter 2. Section 4.5 tests the set of alternative hypotheses that are suggested in the literature as possible causal explanations for variations in the rate of family following between countries. Section 4.6 examines the evidence on two other factors, age and gender.

Section 4.7 is a multivariate regression analysis of the effect of all known and suspected independent variables on the dependent variable, the rate of family following. It seeks to establish which variables impact the rate of family following, the extent of that impact and, by controlling for all variables, separate out the effect, if any, of the electoral system on the rate of family following.

84

4.1 Overall Results

As can be seen from graphs 1 and 2 below, the headline results are in line with what the

Theory of Dynasty Formation would suggest.

Graph 1: Rate of Family Following by National Electoral System 25.00%

20.00%

15.00%

Rate of Family Following by 10.00% Electoral System

5.00%

0.00% CLPR SMP AV (Aus) MMP SMP OLPR PRSTV (Neth) (Can) (NZ) Open (Bel) (Irl) Primaries (USA)

The rate of family following is lowest under systems that do not allow for personalised voting and highest where personalised voting, full intra-party competition and preference vote seeking from other parties are combined in the presence of intra-party competition.

Family following increases, as predicted, as we move from systems that display none of these factors through systems that display some of them to the one system – PR-STV- that

85 displays all of them. Within New Zealand, as predicted the rate of family following among

List MPs is broadly comparable with that seen in the Netherlands while the rate of following among constituency MPs is notably higher.

Graph 2: Rate of Family Following by Electoral System with New Zealand as Two Systems 25.00%

20.00%

15.00%

Rate of Family Following by Electoral System with NZ as Two 10.00% Systems

5.00%

0.00% CLPR MMP SMP AV MMP SMP OP OLPR PRSTV (Neth) List (Can) (Aus) Const (USA) (Bel) (Irl) (NZ) (NZ)

Australia, as predicted, does not show a higher rate of family following than Canada. Indeed at 2.67%, the rate of family following under the more open, Alternative Vote, is actually marginally below the 2.96% recorded under the less open SMP. As detailed below, previous studies on Australia (Lumb 2012; Lumb and Gobbet 2015) have put its rate of family following at 4%, but the different methodology used in these studies accounts for the higher figure, see section 4.6.3 for further details. The rate of parent-child family following in

Australia is at 2.0%, marginally below both that of Canada at 2.07% and, as we have seen

86 above, the UK at 3% and strongly suggests that the degree of electoral system openness alone cannot successfully account for the rate of family following. Looked at in table form and separated by country, the results are as follows:

Table 5: Rate of Family Following by Country

Closed-List Alternative SMP (open Open-List SMP MMP PR-STV PR Vote primary) PR

New United Netherlands Canada Australia Belgium Ireland Zealand States

1.34% 2.96% 2.67% 5.83% 6.21% 14.67% 19.62% i

As expected, with the minor exception of Australia, the rate of family following increases as we move from left to right. Separating the results for New Zealand by electoral system feature gives further support to the proposition that the rate of intra-party competition that is a key causal variable in higher rates of family following.

87

Table 6: Rate of Family Following by Electoral System Features

SMP (open primary)/ Alternative MMP Open-List Closed-List PR SMP Vote Constituency PR PR-STV United Netherlands/ NZ States/New List Canada Australia Zealand Belgium Ireland PV+ FIPC No PV PV PV + ç PV + PIPC PV + FIPC + ç 1.34%/2% 2.96% 2.67% 6.21%/8.45% 14.67% 19.62%

The actual numbers of family followers in each country in the sample, as opposed to their proportion of each parliament, is shown below. The lower houses of parliament in the

Netherlands, Australia, New Zealand, Belgium and Ireland are of approximately equal size and for these countries the graphs for the numbers and percentages of family followers largely mirror each other. The parliaments of the United States and Canada are significantly larger and thus have a higher total number of family followers than the proportionate prevalence of family following in those countries would otherwise suggest. The effect of any bias towards these larger countries should be borne in mind in the sections that follow.

Graph 3: Number of Family followers per country

35 30 25 20 15 10 5 0

88

4.2 Individual Country Findings:

Table 7: Results Summary by Country

New United Netherlands Canada Australia Zealand States Belgium Ireland Rate of Family Following 1.34% 2.96% 2.67% 5.79% 6.21% 14.67% 19.62%

Same Constituency n/a 60% 50% 50% 77.78% 77.27% 80.67%

Direct Inheritance n/a 16.67% 50% 0% 71.43% 18.18% 69.23% Trad Successful Family Parties Followers 100% 100% 75% 85.71% 100% 63.64% 96.67% Non-Followers 62.16% 99.70% 86.30% 74.56% 100% 65.63% 72.65% Family Cabinet Followers 0% 50.00% 75.00% 14.29% n/a 54.55% 25.81% Non-Followers 2.70% 15.55% 24.66% 33.33% n/a 24.22% 18.11% Family Prior Local/Reg exp Followers 0% 20.00% 50.00% 14.29% 51.85% 72.73% 75.76% Non-Followers 50.00% 28.66% 13.69% 16.67% 65.93% 64.84% 80.80% Family Rural Followers n/a 50.00% 50.00% 57.14% 18.52% 0% 75.00% Non-Followers n/a 45.13% 33.56% 20.17% 32.59% 0% 48.80% Family Graduates Followers 100.00% 80.00% 50.00% 57.14% 92.59% 90.91% 70.90% Non-Followers 79.73% 72.56% 75.34% 78.95% 92.89% 89.84% 50.39% Family Social Conservative Followers 100% 30.00% 50.00% 57.14% 37.04% 50.00% 80.65% Non-Followers 50.67% 30.18% 51.33% 60.00% 56.78% 53.33% 59.49% Average Age 1st Family Election Followers 40 41 45 46 49 35 40 Non-Followers 41 45 30 42 47 38 42 Family Women Followers 0% 10.00% 0% 0% 29.63% 18.18% 12.12% Non-Followers 39.19% 26.83% 29.45% 35.96% 17.16% 42.18% 25.60%

The table above shows the percentage of family followers and, where relevant, non-

followers in each country that are coded positive in this study for the attribute indicated.

For example, 60% of family followers in Canada represent the same constituency as their

preceding relative, 100% of family followers in the Netherlands represent traditionally

successful parties compared to 62.16% of non-followers and 50% of family followers in

89

Australia are graduates compared to 75.34% of non-followers. These figures are explained in greater detail below.

4.2.1 Netherlands:

As can be seen in Table 7 above and as predicted by the theory, the rate of family following in the Netherlands is the lowest among the countries being looked at. Just two MPs in the

House of Representatives, Pieter Duisenberg and Pieter Heerma, are closely related to current or former MPs. A third, Hester Maij is the daughter of a former MEP and government minister, but she does not fall within the definition outlined in Chapter 2.

Given the very small sample size great care must be taken not to read too much into these sub-national findings.

Both Dutch family followers are male as are 61.33% of Dutch MPs. Both are graduates, as are 80% of their fellow MPs. Their average age at first election was 40, just below the average for the parliament as a whole of 41. They differ from their peers in not having prior experience of elected office at local or provincial level, again in keeping with expectations.

51% of Dutch MPs have served time in town halls and provincial assemblies before entering parliament, both family followers skipped this stage.

Both represent parties of the centre-right. Both are the sons of prominent former MPs and cabinet ministers and both bear their father’s surnames. Enneus Heerma formerly led the party his son now represents in Parliament. Both of them inherited their seats considerably after their fathers’ retirement from active politics; fifteen years later in the case of Heerma and thirty four years in the case of Duisenberg, though as a minister, of the Dutch

90

Central Bank and later President of the ECB, Wim Duisenberg remained a prominent figure until nine years before his son’s first election in 2012. In neither case though is it possible to speak of the inheritance of a personal political machine, nor should it be expected that this would be the case, such personal machines being unnecessary in the Netherlands.

Given the lack of personal vote seeking in Dutch elections, this should not be surprising. The

1.34% rate of family following in the Netherlands is still much higher than what would exist by mere random chance. It may perhaps be considered the base rate that family socialisation produces.

Of greater significance perhaps is that both are children of former prominent cabinet ministers. Just 2.67% of the members of the current Dutch House of Representatives are former cabinet ministers (none are current), by far the lowest rate of any of the parliamentary systems in this study. The ability to appoint cabinet members from outside parliament and the requirement that any appointed from parliament must resign their seats is surely a factor in this. That the only family followers in the Dutch parliament should be the children of ministers however and that a third MP should be the daughter of a minister who was never an MP does seem to strenghten the idea of family following as a form of elite power perpetuation.

4.2.2 Canada

Although the current Prime Minister Justin Trudeau is the son of former Prime Minister

Pierre Trudeau, political dynasties in Canada are, in fact, relatively rare. Matthew Godwin

91

(2013) put their rate at a steady 3% over the last 50 years, a figure supported by the

Parliament of Canada website. That figure was based on a wider definition of family following than the one used in this study and included among the family followers David

Graham MP, whose great-great-uncle Leon David Crystal was an MP until 1963. Taking the narrower definition used in this study, two degrees of relationship, the rate of family following in Canada comes in at 2.96%; higher certainly than seen in the Netherlands, but, as the theory would predict, among the lower levels observed. As with the Netherlands, the sample size is very small and once again great care should be taken not to read too much into these sub-national findings.

Of the ten family followers in the current Canadian House of Commons, nine are men, with just one woman. This contrasts with the general membership of the House where 27% of

MPs are women. Seven are sons of former MPs, two are grandsons and one is a grand- niece. Neither the grandniece, Chrystia Freeland MP, or the grandsons have the same surname as their predecessor.

60% of family followers represent the same constituency in parliament as their predecessor indicating a strong local element to their succession; though for five out of the six the inheritance was delayed. This suggests that while a local brand name may matter under

SMP in Canada, inheriting a personal political machine is far less important. This is in keeping with the Theory of Dynasty Formation outlined in Chapter 2. The lack of intra-party competition under SMP removes the need for a candidate to construct a private machine separate to that of their parties.

92

Family followers in Canada are noticeably more successful at climbing the greasy poll: half are current or former cabinet ministers, more than three times the 15.6% rate achieved by non-followers. They also enter parliament at a younger age, on average four and a half years before non-followers. The high rate of cabinet promotion persists despite the fact that Canada has equal gender representation at cabinet, something that should militate against the heavily male dominated Canadian political dynasties, but has clearly not done so. Instead it is the chances of promotion for men unrelated to previous office holders that have been constrained.

As Hess (1997) would suggest, family followers are better educated; 80% hold university degrees compared to 73% for non-followers, but this is hardly strong enough to explain their much higher rate of promotion. Fewer family followers have experience as local or provincial elected representatives, with just 20% having served as local councillors or provincial MP compared to 27% of non-followers.

60% of family followers in Canada are members of the left of centre Liberal Party, higher than the 54% of seats held by the party in Parliament and contradicting the idea that conservative parties are more likely to demonstrate family following (Doyle el al. 2015).

50% of family followers represent rural constituencies, higher than the 45.12% found among non-followers and correspondingly 40% of family followers are elected from urban constituencies as opposed to 46.95% of non-followers. While neither figure is remarkably stark, both offer support for the idea that the rate of family following may be affected by varying levels of development within a country.

93

4.2.3 Australia

The rate of family following in Australia according to the methodology used in this study is

2.67%. This is lower than the 4% rate of family relations reported by Lumb and Goddett

(2015). The difference is accounted for by the fact that Lumb and Goddett, like the

Parliament of Canada website and, include relatives beyond the two degrees of relation specified in this study. Australia has been a democracy long enough for an MP like Jane

Prentice to have had a great grandfather who served in Parliament. A second difference is that Lumb and Goddett (2015) is a study of family relations in Parliament, not of family following. Thus where two relatives serve at the same time, Lumb and Goddett count both, while this study only counts the later arrival in Parliament. For example Laurie Ferguson MP was elected to Parliament in 1990. His brother Martin was elected in 1996 and retired in

2013. Lumb and Goddett (2015) list Laurie Ferguson as being related to a previous office holder, this study does not as according to the definition outlined in Chapter 2, Martin, not

Laurie was the family follower. The effect of these methodological and definitional differences is that the Lumb and Goddett figure for the rate of family relationships in the

Australian parliament is consistently higher than the rate for family following produced by this study.

At 2.67% that rate is broadly in line with, if marginally lower, than that found in Canada and thus supports the Theory of Dynasty Formation outlined in this thesis. Australian elections allow preference vote seeking from other parties and are rated as more open than SMP in the literature (Carey and Shugart 1995; Farrell and McAllister 2006, Colomer 2011).

However, they lack intra-party competition and do not incentivise the creation of a personal political machine. As in Canada (Godwin 2013), the rate of family following in Australia has

94 been remarkably consistent for decades (Lumb 2012; Lumb and Godett 2015) and the headline figure may thus be given greater credence. Despite this, the sample size remains very small and caution continues to be advised regarding any conclusions that might be drawn from the sub-national findings.

Family following in Australia is very male dominated. In fact all four current MPs who are closely related to former MPs are men; two sons and two sons-in-law. By contrast 27% of non-family followers in the Australian House of Representatives are women. Both sons directly succeeded their fathers to the same seat upon the fathers’ retirement, thus inheriting their fathers’ political machine. In the case of Bob Katter MP it was a double inheritance. He succeeded to his father’s national seat, while his own son Rob, took his place in the Queensland Parliament. Just 15% of Australian MPs have served time in local or provincial government, but half of the family followers have done so – the same ones who directly inherited their father’s seats. Both of them had served apprenticeships in regional government while their fathers still represented the constituencies in parliament.

As with Canada, Australian family followers are much better represented at cabinet. Three out of the four are current or former Ministers, more than twice the 34% of non-family MPs who have achieved such a promotion. Unlike their Canadian counterparts however, they are less likely to be graduates. Neither of the Australian family followers who are sons of former MPs attended university, though both of those who are son-in-laws of former MPs did. At 50%, the proportion of graduates among Australian family followers is below the

73.4% seen in the rest of Parliament, meaning that in Australia the proposition put forward by Feinstein (2010) and Daniele and Geys (2014) that family followers are less qualified than

95 non-followers receives support. The better promotion prospects enjoyed by Australian family followers despite their markedly inferior educational achievements suggests that, here at least, the hereditary transfer of power is more important than meritocratic achievement.

50% of family followers in Australia represent rural constituencies compared to 33.5% for the rest of Parliament. Correspondingly fewer family followers represent urban constituencies; 25% versus 44.5% for non-followers. On political ideology they are evenly split, two representing the left of centre Labor party, one the right of centre Liberals while the last is a right leaning Independent.

State Level Findings:

At state level, Australia uses two electoral systems: PR-STV in Tasmania and the Alternative

Vote everywhere else. A series of elections took place to Australia’s six state legislatures between 2013 and 2016 returning some 355 parliamentarians. As can be seen from the table below there is limited variation in the rate of family following across four of the five states that use the Alternative Vote; ranging from a low of 3.39% in Western Australia to a high of 5.62% in Queensland. Two of the states show notably higher rates; South Australia, which also uses AV has a family following rate of 8.51% and Tasmania, which uses PR-STV, has an even higher rate of family following of 20%.

96

Table 8: Rate of Family Following in Australian State Legislatures

Election Electoral % Family Province Year System Seats Western Australia 2013 AV 3.39% Victoria 2015 AV 3.41% New South Wales 2015 AV 4.30% Queensland 2015 AV 5.62% South Australia 2014 AV 8.51% Tasmania 2014 PR-STV 20.00%

From the voters’ point of view there is little difference between the electoral systems used on either side of the Tasman Strait. Under both systems a single vote is cast for a candidate.

The voter may express a lower preference or preferences for other candidates to be used in the event that the electors higher preferences do not contribute to the election of a candidate.

Candidates in AV elections therefore, unlike their counterparts running under PR-STV, do not experience intra-party competition and have little incentive to create and maintain a personal political machine. The one exception to this scenario is when former members of a political party run as Independents against their erstwhile party. It is perhaps worth noting that two of the five family followers in the Queensland Legislative Assembly are members of the tiny Katter Australia party and both inherited their seats from relatives who were once elected representatives for other parties.

Candidates for the Tasmanian House of Assembly by contrast experience in full the heightened level of intra-party competition promoted by PR-STV. Indeed, all of the family followers returned to the lower house of the Tasmanian parliament in 2014 were members

97 of political parties who had to compete for election against running mates from their own party.

The findings are thus in keeping with the Theory of Dynasty Formation, but considerable caution is advised. If Tasmania were an independent country it’s small population of just over 500,000 would rank it as a micro-state and therefore lead to its exclusion from this study.

4.2.4 New Zealand

The MMP electoral system used in New Zealand is a hybrid of Closed-List PR and SMP.

Nonetheless both Carey and Shugart (1995) and Farrell and McAllister (2006) regarded it as producing a higher degree of personalised vote seeking than either. The existence of a separate party list gives a greater degree of voter choice, allowing a voter to express a local preference when selecting their constituency representative while still making a party-based choice on a national level. The rate of family following in New Zealand, at 5.79%, seems to bear this out. As outlined in Chapter 3, from a candidate perspective, the manner in which

MMP operates in New Zealand creates, in some cases, a significant incentive to create and maintain a personal political machine, but only for those MPs elected from constituencies and who represent parties who might expect a large List vote.

The 2014 New Zealand election confirms this. Of the seven MPs who are family followers, six are constituency MPs and one was elected on a party list. More MPs in New Zealand are elected on a constituency than on a list basis – 70 to 50 - thus 8.6% of constituency MPs are family followers compared to 2% of List MPs. All six family followers elected for

98 constituencies represent traditionally successful parties, the single family follower elected on a party list is also the only one from a smaller party.

Once again as in the first three countries studied, New Zealand family followers display a striking gender bias. In a parliament where over a third of MPs are women, all seven family followers are men, two sons, a brother a nephew and three grandsons. Four of the seven bear the same surname as their predecessor. Again, as in Canada and Australia, rural areas are over-represented: 57% of family followers represent rural areas in parliament, compared to just over 20% of non-followers. In common with Australia and unlike Canada or the Netherlands, in New Zealand family followers are less well educated than other MPs;

57% are graduates compared to 79% of non-family followers.

This may be a partial explanation for the failure of family followers in New Zealand to achieve cabinet office on anything like the scale of their peers in the other countries.

Uniquely in this study, the proportion of family followers advancing to cabinet is lower than the proportion of non-followers. One third of non-family followers in the New Zealand

Parliament are current or former ministers, compared to just a single family follower: Todd

McClay MP.

Proportionately fewer family followers have held sub-national political office – 14% versus

16% - and they are on average five years older when first elected. Most family following in

New Zealand, as in Canada, is by means of delayed rather than direct inheritance. Though

60% of family followers represent the same constituency as their forebears, none inherited directly on the death, defeat or retirement of their relative. One MP Rino Tirikatene did

99 take over his father’s candidacy when the latter died during an election campaign in 1996, but he was unsuccessful and had to wait another 14 years before winning back his father’s seat. Similarly though three of Adrian Rurawhe’s family held his seat before him, the last one to do so left office in 1969 while he did not enter Parliament until 2014. It is less obvious, therefore than is the case with Australia, that the New Zealand family followers have inherited political machines.

Family followers in New Zealand are less socially conservative than non-followers. While

57.14% of New Zealand family followers represent socially conservative parties, this is marginally less than the 60% of non-followers who are members of such parties.

It should be noted that two of New Zealand’s seven family followers are members of the

Maori nation. The precise implications of this are beyond the scope of this study, but both represent the socially liberal Labour party, both represent rural constituencies, neither are graduates and neither have served in cabinet.

4.2.5 United States

The nature of family following in the US represents a significant change from the first four countries in this study. As the theory would suggest, the rate of family following at 6.21% is higher than in countries with lower levels of intra-party competition. This result is in line with Dal Bó, Dal Bó and Snyder’s (2009) findings of a consistent rate of family following in the US of 6% over the 50 years to 2008. Family following in the US is also much more

100 clearly constituency based. 77.8% of US family followers represent the same constituency as their relative did and 71.5% of these inherited their seat directly.

Because of the larger size of its parliament, the USA has a correspondingly larger number of family followers in absolute terms. This larger sample size means that a little more confidence can be expressed in the findings from the US sample than from those in the

Netherlands, Canada, Australia and New Zealand.

Seventeen of the twenty seven family followers are children, two are grandchildren, four siblings, one niece, one nephew, two widows and a spouse of a Senator who took over his seat upon his elevation to the Upper House. One – Mario Diaz Balart of Florida – is both the brother of a former US Congressman and the son of the former leader of the pre- revolutionary Cuban House of Representatives, a rare example of a transnational political dynasty.

The gender bias in the US House runs in the opposite direction to every other country in this study. While most family followers are still men, women at 30% of the total are significantly better represented among family followers than among non-followers, where they make up just 17% of the members of the House. Family names are still important however. Six of the eight women family followers inherited their seat directly from a male relative. All six use either the same surname as that relative or a double-barrelled name incorporating their male predecessor’s surname. This does occur among some women family followers in the other countries in this study, but to nothing like this level.

101

Almost all members of the US House of Representatives are graduates with 93% of both family followers and non-followers having completed their primary degrees. Family followers are a little younger when first elected, but again the difference is slight – 45 years and 4 months versus 46 years and 9 months.

The biggest differences between family followers are in terms of party affiliation and the nature of the constituencies they represent. The current House of Representatives has a right of centre Republican majority, but 63% of family followers are left of centre

Democrats. Similarly, although the number of rural districts in the US outnumbers the number of urban ones by 31.7% to 30.1%, family followers in the US are much more prevalent in urban constituencies. 51.9% of family followers are elected in urban constituencies compared to just 18.5% who represent rural districts. This urban bias is reinforced by the much greater tendency of urban areas to elect Democrats.

The process of inheritance in the United States; mostly direct and mostly same- constituency, matches that predicted by the theory. The demographic profile of the family followers disputes Hess (1997), as they are no better educated than non-followers; Smith

(2012) as more developed areas are more, not less, likely to return family followers; and

Doyle et al. (2015), because in the US - as in Canada and New Zealand - it is the centre-left party, not the centre right that pre-dominates among family followers.

102

4.2.6 Belgium

As predicted in the theory, the rate of family following in Belgian federal elections using

Open-List PR, at 14.67%, is significantly higher than in electoral systems with lower levels of intra-party competition.

Belgian family followers are disproportionately male to a significant degree. Women make up just 18% of family followers in Belgium despite comprising 44% of non family followers.

The education levels of family followers and non-followers are almost identical, with 90% of both groups having attained a university degree. Once again the ability of family followers to attain cabinet rank is greater than non-followers. 45.5% of family followers are current or former ministers, compared to 25.8% of non-followers. Family followers are elected for the first time on average two and a half years earlier than non-followers. On ideological grounds, Belgian family followers are evenly split: eleven represent parties of the centre right and eleven parties of the centre left. All Belgian constituencies are either urban or mixed, none count as rural. More family followers represent an urban constituency than non-followers: 45.45% to 33.49%.

Family following also occurs, as predicted, predominantly in the same constituency, but a note of caution needs to be sounded here. Belgium uses a small number of large multi-seat constituencies. The country is also deeply divided along language lines, meaning in effect that a Flemish or French speaking politician seeking to stand for election has a maximum of six constituencies in which they might reasonably stand; the Flemish or French speaking areas, and the bi-lingual capital region. The pre-dominantly French speaking province of

Liège is also home to a German speaking minority, but currently all MPs from Liège are

103

French speakers. Under these constricted circumstances, same-constituency following is considerably more likely in Belgium than it would be in a country with a larger number of constituencies and a less pronounced linguistic divide.

A second peculiarity of Belgian politics is the number of layers of government and the tendency of politicians to move up and down between Federal, Provincial and local government, sometimes resigning a seat in the federal parliament to take up a provincial ministry or the mayorality of a city. Thus politicians may remain highly active long after their parliamentary career has apparently ended. As their future career prospects may be affected as much by what happens at a local or provincial level, it also encourages a greater level of involvement in these layers of politics. This may be why, contrary to the experience elsewhere, so many Belgian MPs currently hold local or provincial office, indeed many continue to seek election to local councils and regional parliaments while remaining members of the national parliament. The proportion of MPs with prior experience in local and regional government is higher among family followers in the Chamber of Deputies at

72.7% than among non-followers at 64.8%.

The combined effect of these elements is to make delayed, rather than direct, same- constituency following the dominant form of succession in Belgium. 86.4% of Belgian family followers represent the same constituency as their forebears, but just 21% of those succeeded their relatives directly into parliament.

Because it provides data on relatives who have held local or provincial office, the information from Belgium allows us a glimpse at a possible broader definition of family

104 following to include Members of Parliament whose relatives may have been prominent local or regional figures, but were never elected at a national level. A case in point would be

Stefan Vercamer MP. He does not count as a family follower as defined in this study, yet his grandfather was a local alderman, his father was the mayor, one brother serves as a city councillor and another is the Deputy President of East Flanders, the region Stefan Vercamer represents in parliament. It could well be argued that such a broad range of local and provincial connections are at least as valuable to a candidate’s political career as having a relative in parliament itself.

If this broader definition of family following is used we find that 26.67% of Belgian MPs have close relatives who have held elected office before them. As such detailed information does not exist for our other cases, this – like the state level data in Australia - is provided for illustrative rather than analytical purposes.

4.2.7 Ireland

As predicted by the theory Ireland, at 19.62%, has by far the highest rate of family following among the countries in this sample. As will be shown in the next chapter, 19.62% actually represents something of a low point for Irish political dynasties.

The rate of family following in Ireland was stable from 1981 to 2007 at between 22% and

25%. It declined sharply for the first time ever in 2011 following Ireland’s economic crash hitting 15.1% in that year’s general election. The 2016 figure marks a partial recovery in the fortunes of Ireland’s political dynasties, but remains the second lowest rate of family

105 following since the children of former TDs first emerged as a larger group in Dáil Éireann than veterans of the War of Independence in 1969 (Farrell 1970).

Family following in Ireland occurs overwhelmingly in the same constituency. 83.9% of Irish family followers represent the same Dáil constituency as their predecessor. 69.2% of these inherited their seat directly upon their relative’s departure from office. Like Belgium,

Ireland uses multi-seat constituencies, but the district magnitude is significantly smaller ranging from just three to five seats. The chance of two relatives representing the same constituency by random chance is thus far less. It is clear then, that in most cases political inheritance in Ireland consists of the inheritance of a functioning political machine.

Women make up just 12.9% of Irish family followers, a much lower share than their 25.6% of non-family seats. This is the reverse of the situation pertaining from 1981 to 2007, when women were more strongly represented among the political dynasties. In 2011 there was a complete wipe out of women family followers, with every single female member of Ireland’s political dynasties losing their seat. At the same election the overall number of women in the Dáil increased slightly, masking a more significant increase in the number of women from non-dynastic backgrounds. 2016 marks a partial recovery for dynastic women, but this coincided with the introduction of candidate gender quotas for political parties, an initiative which has led to a 40% increase in the overall number of women in Dáil Éireann.

The majority of Irish TDs spend a period of time as elected councillors in local government prior to entering the Dáil, to raise their profile and help build a local political base. Up to

2003, as is still the case in Belgium, most TDs also served on local authorities. From 1981 to

106

2003 family followers were significantly less likely than non-followers to have been councillors before their election to the Dáil, indicating a lesser need to build such a base (see

Chapter 5). The abolition of the dual mandate in 2003, when sitting TDs were compelled to resign their local authority seats, led to a sharp reduction in this gap over the next four elections. TDs who were sitting councillors had to resign their seats on local authorities.

These seats were filled by co-option by the parties locally with the TDs chosen successor usually getting the job. In many cases this proved to be a family member who went on to contest for a Dáil seat in subsequent general elections. This gap in local authority experience has now effectively disappeared. In the current Dáil 74.2%, of family followers have served previously as councillors compared to 74.8% of non-followers.

There is a clear party bias amongst Irish family followers. 26% of the centre right Fine Gael party’s TDs are closely related to previous TDs or Senators as are 25% of the ideologically similar Fianna Fail party. The centre left Labour party’s rate is even higher at 28.6%, though this is most likely related to the party’s very poor performance in 2016 and the skewing effects of small sample sizes. At the previous election in 2011, Labour secured 2nd place with 37 seats. Just 10.8% of Labour TDs in 2011 were members of political dynasties. In

2016 Labour lost more than three quarters of it’s seats. Labour’s family followers also suffered, losing half their representation, but in proportionate terms the share of the now much reduced Labour parliamentary party that is made up of political dynasties rose.

Parties of the radical left have significantly fewer family followers. This may be down to ideology as Doyle et al. (2015) suggest, but it is more likely proof of the third principle outlined in the Theory of Dynasty Formation put forward in Chapter 2:

107

 Dynasties take time to form.

The far left AAA/PBP, now renamed Solidarity, has no family followers, but the party only entered Dáil Éireann in 2011, as the United Left Alliance, leaving little time for dynasties to form. The nationalist/left Sinn Féin party has been in the Dáil since 1997, but only rose above single figures for the first time in 2011. A number of Sinn Féin dynasties have emerged at local level, but just one TD, who was married to a sitting

Senator at the time of her election, counts as a family follower.

Ireland, in common with Canada, Australia and New Zealand, displays a rural bias towards political dynasties. 75% of family followers represent rural areas, compared to just 48.8% of non-followers. Also as we have seen in every country except New Zealand, Irish family followers are more successful at attaining ministerial office than their non-dynastic peers.

25.8% of family followers in Dáil Éireann are current or former cabinet ministers, compared to 18.1% of non-followers.

At least a partial explanation for this may be their level of education. Ireland is the only country in the survey to give strong support to the Hess (1997) notion that family followers enjoy greater success because they are better educated than non-followers. Just 50.4% of non-family followers in the Irish Dáil hold a university degree. The next lowest is Canada at

72.6%. The proportion of graduates among family followers at 70.9% is at the mid-range of this study, but as family followers are a minority in each of the parliaments studied the overall proportion of graduates in the Irish Dáil is, at 54.4% substantially lower than any of the other countries in the study, all of which have rates above 72%. Why Irish politicians

108 should be so poorly educated by comparison with their peers is beyond the scope of this work, but is perhaps something that merits further study.

4.3 Detailed Hypotheses:

The Theory of Dynasty Formation posited in Chapter 2 proposed that variations in the rate of family following between mature, developed, democratic states are a function of the incentive to create and maintain a personal political machine and that therefore states which use electoral systems that reward the creation and maintenance of such personal political machines will see higher rates of family following. The headline figures shown above strongly support this central proposition.

The theory further suggests that the incentive to create and maintain a personal political machine is affected by a number of elements of the electoral system namely, – personalised voting, intra-party competition and preference vote seeking from other parties in the presence of intra-party competition. These individual elements gave rise to a number of detailed hypotheses that will now be tested further:

Hypothesis 1: Political dynasties will be more prevalent in electoral

systems that allow for personalised voting

All electoral systems in this study other than the Closed-List PR system used in the

Netherlands and the List element of the MMP system used in New Zealand allow for personalised voting and are thus held to have some degree of incentive towards the

109 creation and maintenance of a personal political machine. 1,303 MPs in this study were elected under systems that allow for personalised voting, 99 of whom are family followers.

By contrast 199 MPs were elected under systems that do not allow for personalised voting,

3 of whom are family followers.

Comparing the New Zealand List and Dutch MPs with those elected directly under the other systems in this study produces we can see that, in line with the Theory of Dynasty

Formation, the rate of family following in systems that allow for Personalised Voting is significantly higher - more than five times higher - than in those that don’t.

Graph 4: Effect of Personalised Voting on rate of Family Following

8.00%

7.00%

6.00%

5.00%

4.00%

3.00%

2.00%

1.00%

0.00% Non Personalised Voting (Neth & NZ List) Personalised Voting (All others)

Running a z test comparing the proportion of family followers elected under non- personalised voting systems with the proportion of family followers elected under personalised voting systems, we find that non-personalised voting is quite heavily negatively correlated with family following. The results of this test are statistically significant.

110

Table 9: Personalised Voting Z Test

Z score P value Non-Personalised v Personalised voting -3.18 0.001

Thus Hypothesis 1 that Political Dynasties will be more prevalent in electoral systems that allow for Personalised Voting is thus supported by the evidence.

Hypothesis 2: Electoral systems that allow for intra-party competition

will have much higher rates of family following than those that do not

Intra-party competition exists, in part, under the SMP (open primary) system used in the

United States. As outlined in Chapter 3, Constituency MPs in New Zealand may also find themselves in a situation where they are running against their own party, while candidates in Belgium and Ireland face full-throated intra-party competition.

Comparing the Belgian, Irish, American and New Zealand Constituency MPs with those elected under systems devoid of intra-party competition results as follows:

111

Graph 5: Effect of Intra-party Competition on rate of Family Following

12.00%

10.00%

8.00%

6.00%

4.00%

2.00%

0.00% No Intra-party competition Intra- Party competition (Neth, Can, Aus NZ List) (US, NZ Const, Bel, Ire)

Of the 688 MPs elected under systems that do not allow for intra-party competition, 17, or

2.47% are family followers. Of the 814 MPs elected under systems allowing either partial or full intra-party competition 86, or 10.57% are family followers.

A z test on these figures shows, as expected, a very strong positive correlation between intra-party competition and family following. That correlation is statistically significant.

Table 10: Intra-party Competition Significance Test

Z score P value Intra-Party Competition v no Intra-Party Competition 6.184 0

When the figures for partial and full intra-party competition are examined separately the effect of the incentive to create and maintain a personal political machine becomes even clearer.

112

Graph 6: Effect of Partial and Full Intra-party Competition on rate of Family Following

20.00% 18.00% 16.00% 14.00% 12.00% 10.00% 8.00% 6.00% 4.00% 2.00% 0.00% No Intra-party competition Partial Intra-party competition Full Intra-party competition (Neth, Can, Aus, NZ List) (USA & NZ Constituency MPs) (Belgium & Ireland)

Of the 506 MPs elected under systems allowing for partial intra-party competition – SMP

(open primary) and the constituency element of MMP, 33, or 6.53% are family followers. Of the 308 MPs elected under systems allowing for full intra-party competition 53, or 17.21% are family followers.

Hypothesis 2 Electoral systems that allow for intra-party competition will have much higher rates of family following than those that do not is again supported by the evidence.

Hypothesis 3: Electoral systems that allow for preference vote seeking

from other parties but do not allow for intra-party competition will not

demonstrate notably higher rates of family following.

113

The seeking of preference votes from competitors outside a candidate’s own party is a feature of Irish and Australian politics. Looked at in the aggregate there appears to be an effect, but in this case it is the very high rate of family following in Ireland that skews the result. As has been seen, the AV system in Australia actually produces a marginally lower rate of family following than SMP in Canada and the UK, suggesting that preference vote seeking from other parties without intra-party competition has little verifiable impact on the rate of family following.

Graph 7: Effect of Preference Vote Seeking from Other Parties on rate of Family Following

12.00%

10.00%

8.00%

6.00%

4.00%

2.00%

0.00% Without Preference Vote Seeking from With Preference Vote Seeking from Other Parties (All except Irl & Aus) Other Parties (Irl & Aus)

When the effect of intra-party competition is separated out the results are stark. A TD in

Ireland, where the electoral system combines preference vote seeking from other parties with intra-party competition is nine times more likely to be a family follower than an MP in

Australia where the electoral system features preference vote seeking from other parties without intra-party competition:

114

Graph 8: Effect of Preference Vote Seeking with & without Intra-Party Competition

20.00% 18.00% 16.00% 14.00% 12.00% 10.00% 8.00% 6.00% 4.00% 2.00% 0.00% Without Preference Vote Preference Vote Seeking Preference Vote Seeking+ Seeking (all except Ireland & without Intra Party Intra-Party Competition Australia) Competition (Australia) (Ireland)

Hypothesis 3: Electoral systems that allow for preference vote seeking from other parties but do not allow for intra-party competition will not demonstrate notably higher rates of family following is held to be supported. Preference vote seeking from other parties without intra-party competition does not significantly increase the incentive to create and maintain a personal political machine over and above the incentive already present under

SMP.

Preference vote seeking from other parties with intra-party competition does increase such an incentive and is correlated with an increase in the rate of family following.

Hypothesis 4: By contrast, electoral systems that allow for intra-party

competition and preference vote seeking from other parties will will

show the highest rate of all

115

Hypothesis 4 is in effect a corollary of Hypothesis 3. As shown above Ireland, which combines both elements, along with personalised voting, has the highest rate of family following by a significant margin. As was argued in Chapter 2 preference vote seeking from other parties with intra-party competition may introduce an enhanced personalised element which can complicate transfer patterns leading to a greater sense of “every candidate for themselves” and thereby adding yet further to the incentive to create and maintain a personal, political machine. As will be seen in the chapter on Ireland, one of the features of PR-STV, and indeed of PR-SNTV in the past, is a high number of Independent candidates. Many of these candidates are former members of political parties and compete directly for the voters of their former parties. In this they act like unofficial additional running mates, but running mates beyond the reach of party discipline or restraint, adding yet further to the already high levels of intra-party competition.

4.4 Additional Observable Implications:

As outlined in Chapter 2, if the Theory of Dynasty Formation holds true it should give rise to a number of additional observable implications:

Observable Implication 1: That because what is being inherited is a local political machine, it follows that instances of same-constituency family following will be higher under electoral systems that incentivise the creation and maintenance of such machines than under those that do not.

116

Graph 9: Effect of Intra-party Competition on rate of Same Constituency Inheritance

90.00% 80.00% 70.00% 60.00% Same Constituency 50.00% Different Constituency 40.00% 30.00% 20.00% 10.00% 0.00% No Intra Party Competition Intra-Party Competition (n=14 cases) (n = 85 cases)

To an extent all politics is local and members of the same family even in developed states tend to live in the same broad geographical areas (Lin and Rogerson 1995). It should therefore be no surprise that most family following across all seven countries in this study occurs on a same-constituency basis. There is however, as would be expected from the

Theory of Dynasty Formation, a markedly higher tendency for same-constituency family following under electoral systems that foster intra-party competition and thereby incentivise the creation and maintenance of personal political machines. Personal political machines are determinedly local entities. While the child of a prominent politician might have good contacts in party headquarters that could secure a nomination in a different constituency, perhaps to a safer seat than that held by their predecessor; the inheritor of a personal political machine is far more tied to the locality of their preceding relative.

Three of the family followers in this study were elected from party lists, two Dutch MPs and one New Zealand List MP, and therefore do not have a constituency. These have been excluded from this and other constituency based observations. Fourteen others were elected under systems that do not foster intra-party competition. Of these, eight, or 57.14%

117 represent the same constituency as their relative did. Eighty five family followers were elected under systems allowing for Partial or Full intra-party competition. Of these sixty seven, or 78.8% hold the same seat once held by a close relative. Thus the gap between the rate of same-constituency inheritance and the rate of extra-constituency inheritance in electoral systems that do not promote intra-party competition is 14.28 percentage points while the gap between the rate of same-constituency inheritance and extra-constituency inheritance in systems that do promote intra-party competition is a far more pronounced

55.8 percentage points.

Once again, the position becomes even clearer when systems allowing partial and full intra- party competition are treated separately. The aggregate rate of same-constituency inheritance in the USA and among constituency MPs in New Zealand, systems classified in this study as being subject to partial intra-party competition, is 72.22%. The aggregate figure for those electoral systems subject to full intra-party competition, Ireland and

Belgium, is a notably higher 81.13%.

As we have seen in the specific country notes in section 4.1, Belgium, because of the large magnitude of it’s constituencies and the constraining effects of the language divide within the country, has a low number of effective constituencies in which a prospective MP might stand for election. This could be expected to make same-constituency inheritance there more likely. It is worth noting then that the rate of same-constituency following in Ireland, where such constraints do not apply, is, at 83.87%, even higher than in Belgium (see graph

10 below). The evidence therefore supports the predicted observable implication that:

118 instances of family following under systems that incentivise the creation and maintenance of a personal political machine will tend to be in the same constituency.

Graph 10: Effect of Full IPC & Preference Vote Seeking on rate of Same Constituency Inheritance 90.00% 85.00% 80.00% 75.00% 70.00% 65.00% 60.00% 55.00% 50.00% 45.00% 40.00% No Intra-party Partial Intra-party Full Intra-party Full Intra-party competition (Can, Aus) competition competition (Belgium) competition & Preference n=12 (USA, NZ Const) n=22 Vote Seeking (Ireland) n=33 n=30

Observable Implication 2: Similarly instances of family following under systems that incentivise the creation and maintenance of personal political machines will tend to happen immediately following the death, defeat or retirement of the preceding family member before the machine has time to decay. Such direct inheritance will be less common under electoral systems that do not so incentivise the creation and maintenance of personal political machines.

A machine that is left inactive corrodes. Clients attach themselves to other candidates, supporters, funders and workers drift away. The longer the gap between one member of a dynasty stepping down and another replacing them, the less of a functional machine there is to inherit.

119

Of the eight family followers elected to the same constituency as their preceding relative under systems without intra-party competition only three or 37.5% succeeded their relative directly upon the latter’s death, defeat or retirement, compared to 62.5% who inherited the seat at a later date. Under systems with intra-party competition by contrast, direct inheritance is notably higher. Some thirty seven out of sixty seven or 55.22% of the same- constituency family followers in this study elected under such systems inherited their seat directly.

Graph 11: Effect of Intra-party Competition on rate of Direct Inheritance

70.00%

60.00%

50.00% Direct Inheritance 40.00% Indirect Inheritance

30.00%

20.00%

10.00%

0.00% No Intra-party Competition Intra-party Competition (n=8 cases) (n=67 cases)

Even this might be considered a little low and the explanation for that lies in the nature of political careers in Belgium. It is commonplace for Belgian MPs to step down from parliament to take up senior roles in the country’s powerful regional assemblies or as mayors of major cities. In such circumstances, they remain active in electoral politics and

120 their personal political machines continue to function in many cases for years after their parliamentary careers have ended. The multi-level nature of Belgian political careers means that the rate of direct same-constituency following in that country is a relatively low 51.16%.

Graph 12: Direct Inheritance by intra-party competition separating Belgium & Ireland 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% No Intra-party Partial Intra-party Belgium Ireland competition competition n=22 n=30 (Can, Aus) (USA, NZ Const) n=12 n=33

By contrast, in Ireland, which lacks powerful regional and local government, the proportion of direct, same-constituency following is 69.23%, while the aggregate figure for the USA and

New Zealand is 62.5%. All three rates are much higher than those found in Canada or

Australia where intra-party competition is not a feature of the electoral system.

The evidence therefore supports the predicted observable implication that: instances of family following under systems that incentivise the creation and maintenance of a personal political machine will tend to happen immediately following the death, defeat, or retirement, of the preceding family member.

121

Observable Implication 3: That in electoral systems that allow for intra-party competition, family following will be more prevalent among more traditionally successful parties, as such parties are likely to evidence higher rates of intra-party competition. This should also hold true for former members of such parties who choose to run as Independents.

The vast majority of MPs in this study - some 1,296 out of 1,502 or 86.28% - represent traditionally successful parties. Overall family followers are more likely to represent such parties than non-followers by a small margin; 89.21% to 86.07%

If the focus is solely on those electoral systems that feature partial or full intra-party competition, namely USA, New Zealand Constituency, Belgium and Ireland, the picture is similar. Some 636 MPs out of 729 non-family followers or 87.24% represent traditionally successful parties. Among family followers elected under these systems, the proportion is higher – seventy six out of eighty five, or 89.41% - but again the effect is limited.

It should be noted however that in the United States, 100% of Members of Congress and therefore 100% of family followers, represent traditionally successful parties. The situation is similar in Canada, where 99.7% of MPs and 100% of family followers represent traditionally successful parties. Given the size of the US House of Representatives and the

Canadian House of Commons relative to the other parliaments in this study, this unusual traditional party dominance skews the figures somewhat. If the USA and Canada are excluded – one from the group of systems without Intra-party competition and the other from the group of systems with intra-party competition - the results become much clearer.

122

Without the US House of Representatives and the Canadian House of Commons the proportion of non family following MPs elected under systems allowing for intra-party competition who represent traditionally successful parties falls to 228 out of 321 or 71.02%, while the proportion of family followers elected under these systems who represent those same parties remains notably higher at forty nine out of fifty eight or 84.48% a statistically significant gap of more than 13 percentage points.

By contrast the figure for those MPs elected under systems that do not allow for intra-party competition, Netherlands, Australia and List MPs from New Zealand, there is really no gap at all. Some 243 of the 350 non-family following MPs in this study elected under these systems or 69.43% represent traditionally successful parties. Among family followers elected under these systems the proportion representing traditionally successful parties is five out of seven or 71.43%, a very minor difference indeed in the context of the very low sample size of family follower in these cases.

As can be seen from the graph below, the predicted observable implication that in electoral systems that allow for intra-party competition, family following will be more prevalent among more traditionally successful parties while present in all cases becomes very clear once the lack of non-traditionally successful parties in the US Congress and Canadian House of Commons is controlled for.

123

Graph 13: % of Family Follower & Non Followers who represent Traditionally Successful Parties by intra-party competition (excluding USA & Canada)

90.00%

85.00%

80.00% Family Followers Non-Followers 75.00%

70.00%

65.00%

60.00% No Intra-Party Competition Intra-Party Competition (Neth, Aus, NZ List) (NZ Const, Bel, Irl)

Observable Implication 4: In countries with mixed systems such as MMP, the rate of family following among constituency MPs will be higher than among list MPs.

The sample size for this measure is very small and accordingly a high degree of caution is advised. However, the predicted implication is easily observed. The rate of family following among constituency MPs in New Zealand is, at 8.45% more than four times that found among List MPs.

Observable Implication 5: That in electoral systems that allow for intra-party competition family followers will be disproportionately represented among high office holders such as cabinet ministers.

124

For this observation, it is again necessary to control for the USA and also for the

Netherlands. The US, uniquely among the countries in this study has a Presidential rather than a Parliamentary system of government. While members of Congress are appointed to cabinet from time to time, it is a comparatively rarer occurence than in other juristictions.

For a member of Congress, election as Speaker or to the chair of a powerful committee are the more likely career goals. These achievements are not directly comparable with the other countries in the study, so for this measure, the US has again been excluded. The

Dutch also appoint cabinet ministers from outside parliament and require all ministers to resign their seats. Currently just 2.7% of Dutch MPs are current or former Cabinet

Ministers, a figure far below that seen in the other five countries in our study.

Excluding the USA and the Netherlands some 191 MPs out of 844 non family followers or

22.63% are current or former cabinet ministers. Among family followers that figure is twenty nine out of seventy five or 39.72% a notably higher promotion rate and clear evidence, across all the electoral systems that being related to a former office holder is good for one’s career. The superior career prospects enjoyed by family followers in so many countries cannot be adequately explained by other factors and speaks strongly of elite power retention and perpetuation. It is this factor above all else that makes the study of family following more than a curiousity. The ability of family followers to achieve high office much more easily suggests that they inherit significant advantages over non-followers. That advantage may also include safer parliamentary seats. Klein and Umit (2016) found, that in the UK at least, Ministers tend to be drawn from those MPs who enjoy the safest seats.

125

The composition of the cabinets in the five countries being looked at in this section strongly supports Observable Implication 5. Of the 115 MPs elected in Canada, Australia and from party lists in New Zealand who are current or former cabinet ministers, just seven, or 6.09% are family followers. Of the 104 current and former Ministers elected in Belgium, Ireland and from constituencies in New Zealand, fully twenty one or 20.19% are family followers.

It is worth noting that at the top of the political ladder, the proportion of family followers is even greater. At the time of their last general election, three of the seven countries in our study, Canada, Belgium and Ireland, were led by family followers; Justin Trudeau son of

Pierre Trudeau, son of Louis Michel and Enda Kenny son of Henry Kenny; meaning that although only 6.79% of the combined members of parliament for each of the seven countries is currently a family follower, the rate of family following among heads of government in our sample is 42.85%.

126

A caveat must be raised with regard to Observable Implication 5. While, as predicted, the overall proportion of family followers in the cabinets of countries whose electoral systems foster intra-party competition is much higher than in those that don’t, the chance of an individual family follower being promoted to cabinet is actually higher in the latter set of countries. Among MPs elected in Canada, Australia and from party lists in New Zealand some 108 out of the 523 non-family followers or 20.65% are current or former cabinet ministers. Among the very small number of family followers elected under these electoral systems, seven out of fifteen or 47.06% have been promoted to cabinet at some point in their careers. Once again the small sample size places a question mark over the statistical significance of these findings.

By contrast some 83 out of 321, or 25.86%, of the non-family followers elected in Belgium,

Ireland and from constituencies in New Zealand are current or former cabinet ministers.

Family followers still fare significantly better than non-followers. Of the fifty eight family followers elected Belgium, Ireland and from constituencies in New Zealand, twenty one, or

36.21%, have been promoted to cabinet, but the advantage is, perhaps, lower than expected.

As discussed in the literature review, Smith and Martin (2017) in their study of family following in Ireland found that the relatives of ministers enjoyed a significant advantage in promotion to cabinet, not just over non-family followers, but also over family followers whose predecessors had not been ministers. They put this down to the enhanced

“information advantage” enjoyed by the relatives of cabinet ministers. This concept is

127 entirely compatible with the machine-based Theory of Dynasty Formation outlined in this study as such an “information advantage”, including such things as contacts in party headquarters, advice and mentoring are part of a family follower’s inheritance. In that sense the child of a cabinet minister embarking on a political career might be held to have inherited a more sophisticated machine than the child of a mere backbencher. The evidence from this study provides support for the Smith and Martin (2017) thesis. Across the five countries looked at in this section some 43.5% of the children of cabinet ministers who enter parliament are current or former cabinet ministers themselves compared to just

38% of those family followers whose predecessor was not at some point a cabinet minister.

In this regard the relatively poorer promotion prospects of family followers in countries whose electoral system allows for intra-party competition may be a function of the much higher numbers of family followers to be found in such parliaments. Smith (2012) found that Japanese family followers, unlike those elsewhere, did not enjoy a deterrent effect against potential challengers, something he put down to the sheer prevalence of family followers in Japan. It may be that where family followers already occupy a disproportionately high number of cabinet seats, the promotional benefit accruing to a new follower, while still present, is naturally diminished.

128

Graph 15: % of MPs who are current or former Cabinet Ministers

50.00%

45.00%

40.00%

35.00%

30.00%

25.00%

20.00%

15.00%

10.00% Non-Followers No Family Followers No Non Followers Intra- Family Followers Intra-party Intra-party Party Competition Intra-Party Competition Competition Competition

Observable Implication 6: The proportion of MPs with prior experience in elected office at local or regional level will increase as the the degree of intra-party competition increases:

The Theory of Dynasty Formation outlined in Chapter 2 suggests that in systems without intra-party competition; Closed-List PR, SMP and MMP List, fewer politicians will serve an apprenticeship at local or regional level before seeking national office, the electoral machine that the party puts at their disposal making such toil unnecessary. In systems with higher levels of intra-party competition we should expect that the need to build a personal political machine will lead to a larger number of politicians overall beginning their careers at local level. A family follower may of course inherit such a machine, making a career at local level unnecessary, but the imperative to protect the family’s base from potential local level

129 challangers should also cause larger numbers of family followers to begin their careers at sub-national level in systems with higher rates of intra-party competition, though to a slightly lesser degree than non-followers.

As we can see from the graph below, the evidence strongly supports this interpretation.

Prior service at sub-national level is highest among all MPs in systems with greater degrees of intra-party competition and lowest in those system where such competition is absent.

Family followers across the board are less likely to serve an apprenticeship at local or regional level than non-followers. The predicted observable implications is clearly present.

Graph 16: MPs holding Prior Local/Regional Elected Office

80.00%

70.00%

60.00%

50.00% Family Followers 40.00% Non Followers

30.00%

20.00%

10.00%

0.00% No IPC Partial IPC Full IPC (Neth, Can, Aus, NZ List) (USA, NZ Const) (Bel, Irl)

130

In conclusion therefore the evidence in support of the Theory of Dynasty Formation outlined in Chapter 2 is considerable. All four posited hypotheses are strongly supported by the results of this study as are all 6 additional observable implications. The sixth implication is partly supported. There is a slight caveat on Observable Implication 5 in that even though family followers as a group make up a much larger proportion of the cabinets of countries using electoral systems that allow for intra-party competition, individual family followers do not appear to be more likely to be promoted to cabinet in these countries.

4.5 Alternative Hypotheses:

There are a number of alternative explanations which give rise to further testable hypotheses:

Alternative Hypothesis 1: That family following, even in mature

developed, democratic states is linked to rent seeking behaviour and that

therefore states with higher perceived rates of corruption.

Transparency International’s Corruption Perception index findings for the countries in this study at the time of the relevant election are outlined in the table below. The higher the score the less corrupt the country is perceived to be. If the hypothesis that family following is linked to rent seeking is correct, then family following should increase from left to right as perceived levels of rent seeking r increase.

131

r

Table 11: Family Following and Corruption Perceptions Index

Country United New Zealand Netherlands Canada Australia Belgium States Ireland TI CPI Score 91 84 83 81 76 74 73 % Family Followers 5.93% 1.34% 2.96% 2.67% 14.67% 6.21% 19.62% f

Thus New Zealand should see a lower level of family following than Canada, the US a higher

level than Belgium, the opposite of what the personal vote seeking hypothesis would

suggest. Ireland would be highest under either hypothesis. For this hypothesis the figure

used for family following in New Zealand is the national one and combines both list and

constituency seats.

As we will see in the multivariate analysis there is a positive correlation between the rate of

perceived corruption in a given country and its rate of family following, however that

correlation is not statistically significant. As shown in the graph below the correlation does

not hold true in all cases, principally because New Zealand’s rate of family following exceeds

that seen in the Netherlands, Canada and Australia, despite its better ranking on perceived

corruption.

132

Graph 17: Rent Seeking & Family Following

100 90 80 70 60 50 TI CPI Score 40 % Family Followers 30 20 10 0

However it should be noted that the four countries in the cross-national study with the lowest rates of perceived corruption are also the four with the lowest rate of family following. Thus while the hypothesis that a causal relationship between higher levels of perceived corruption and the rate of family following in mature, developed, democratic states is not fully supported, a correlation between higher rates of family following and poor governance may indeed exist. The causal relationship though, as the business literature on family following argues, is likely to be the other way around. Rather than corruption leading to high rates of family following, high rates of family following may simply lead to poorer governance and oversight.

133

Alternative Hypothesis 2: That family following, even in mature

developed, democratic states is linked to underdevelopment

The table and graph below lists the countries in our sample from left to right according to

their ranking on the United Nations Human Development Index in the year of the election

included in this study; 2012 for the Netherlands, 2013 for Australia, 2014 for Belgium,

United States and New Zealand and 2016 for Ireland. It can be argued that like corruption,

development is a continuum and that even among mature, developed democracies, some

have attained a higher level of development d than others. If there is a link between

family following and relative development in developed countries, then the rate of family

following should increase as the rate of development falls as one moves from left to right

across the table. If true Canada and New Zealand should evidence a very similar level of

family following, while the rate in Ireland should be lower than either.

Table 12: Family Following and Human Development Index

Australia Ireland Netherlands United States Canada New Zealand Belgium

UN HDI Score 0.936 0.923 0.922 0.914 0.913 0.910 0.881 % Family Followers 2.67% 19.62% 1.34% 6.21% 2.96% 5.93% 14.67% d

134

Graph 18: Family Following & UNHDI

1 0.9 0.8 0.7 0.6 0.5 UN HDI Score 0.4 % Family Followers 0.3 0.2 0.1 0

As will be seen in the multivariate analysis there is a negative correlation between

Development and Family Following, but it is very far indeed from being statistically significant. Looking on a country basis, the observed rate of family following across the sample countries simply does not correlate with their UNHDI rankings. This hypothesis too is not supported.

Alternative Hypothesis 3: That family following, even within mature

developed, democratic states is linked to underdevelopment and thus

will be more prevalent in less urbanised areas.

Smith (2012) found that in Japan family following is more prevalent in rural areas. While as we have seen there is no evidence that variations in the rate of development between

135 mature, developed, democratic states are a causal factor in the rate of family following, it is arguable that relative levels of development within states might have an influence. If true this could in part explain why family following has continued to persist at such consistent levels. Relative differences between levels of development of urban and rural areas within a country are likely to be locked in for the long term. While a middle income country such as

Estonia might become significantly wealthier in decades to come, it is harder to envisage a set of circumstances in which the current difference in the relative levels of development within Australia, say between the Sydney region and the Outback, might dissipate.

As can be seen below, there is indeed a rural bias among family follower on an aggregate level, but unfortunately for this alternative hypothesis, it is relatively minor. The vast majority of MPs representing both urban and rural constituencies are not family followers.

Graph 19: Rate of Family Following in Urban, Rural & Mixed Constituencies

9.00% 8.00% 7.00% 6.00% Family Followers 5.00% 4.00% 3.00% 2.00% 1.00% 0.00% Urban Rural Mixed

Just 37 out of 456 or 8.11% of the MPs elected to represent rural areas, that is constituencies with an overall population density of less than 150 persons per square

136 kilometre and no single settlement of more than 50,000 people, are family followers. This is proportionately a little higher than the 36 out of 498 or 7.22% of MPs representing urban constituencies, that is constituencies with a population density of over 500 persons per square kilometre, who are members of political dynasties, but the difference is not large.

Mixed constituencies show the lowest rate of family following. These constituencies have a populaton density of between 150 and 500 persons per square kilometre or have a population density below 500 per square kilometre but with a single settlement of more than 50,000 people. Just 26 of the 448 MPs elected for mixed areas, or 5.8% are family followers. As can be seen belowm family following does have a rural bias, but it is a relatively modest one and is not statistically significant.

Table 13: Rural z Test

Z score P value Greater likelihood that family followers will represent rural constituencies 1.510 0.131

For alternative hypothesis 3 to hold as an explanation for variations in the rate of family following between countries, however, there would need to be a clear correlation between the rate of family following in a country and the proportion of MPs in its parliament who represent rural constituencies. As we can see from Graph 20 below, this is simply not the case.

137

Though Ireland, which is the most rural country in this study does have the highest rate of family following, the second most rural, Canada, has the second lowest, while Belgium manages to achieve 2nd highest rate of family following with no rural constituencies at all.

Graph 20: Rates of Family Following & Proportion of MPs from Rural Constituencies

60.00%

50.00%

40.00% Proportion of Parliament that is Rural 30.00% Rate of Family Following 20.00%

10.00%

0.00% Ireland Canada NZ Australia USA Belgium (Const)

Alternative Hypothesis 3 is found not to be supported.

Alternative Hypothesis 4: That the rate of family following is simply a

function of the superior quality of candidates from political families, in

particular their higher levels of education

138

Graph 21: Graduates 100.00%

90.00%

80.00%

70.00%

60.00%

50.00%

40.00%

30.00%

20.00% Family Followers Non Followers

As can clearly be seen from the graph above, Hess’s (1997) proposition that family followers are better educated that non-followers and thus Alternative Hypothesis 4: That the rate of family following is simply a function of the superior quality of candidates from political families, in particular their higher levels of education receives very little support from the evidence. No significant difference exists at an aggregate level. Family followers enjoy a slight educational advantage, 81.37% to 79.5%, but it is by no means sufficient to explain their greater propensity to be elected and promoted. Feinstein (2010) and Daniele and Geys

(2014) claim that family followers are less qualified than non-followers are not supported either. Once aggregated, both groups enjoy very similar levels of educational attainment.

A z test reveals shows a minor tendency towards family followers being graduates, but the finding is not statistically significant.

Table 14: Graduates z Test

Z score P value Family Followers v Non-followers 0.249 0.803

139

Alternative Hypothesis 5: That the rate of family following is linked to

traditionalism and will therefore be higher among socially conservative

parties regardless of the electoral system.

Some 747 Members of Parliament, or 49.73% of the total are coded as representing socially conservative parties. Among family followers, some 57 elected members out of 102 or

55.88% represent socially conservative parties. This gap is significant and appears to support the hypothesis. The gap however is entirely explicable by the political dominance of the socially conservative parties in Ireland, Fianna Fáil and Fine Gael. If Ireland is excluded the proportion of family followers representing socially conservative parties in the other six countries falls to 40.27%, below the rate for non-followers of 50.21%.

Moreover for this hypothesis to hold as an explanation for variations in the rate of family following between countries, then, as with Alternative Hypothesis 3, there would need to be a correlation between the proportion of MPs in each parliament who represent socially conservative parties and the rate of family following in those parliaments.

140

Graph 22: Socially Conservative MPs & Family Following

70.00%

60.00%

50.00%

40.00% Proportion of Socially Conservative MPs 30.00% Rate of Family Following 20.00%

10.00%

0.00%

As Graph 22 above makes clear, this is not the case. The data from Ireland, as with

Alternative Hypothesis 3, does seem to support the idea, but across all seven countries there is simply no correlation between the electoral strength of socially conservative parties and the rate of family following. Alternative Hypothesis 5 is not supported.

Alternative Hypothesis 6: That the rate of family following is a function of

the degree of electoral system openness and not of intra-party

competition.

The Alternative Vote system used in Australia is categorised as more Open than the Mixed

Member Plurality system used in New Zealand (Farrell and McAlister 2006; Colomer 2011).

Despite this the rate of family following in New Zealand in this study is, at 5.79%, more than

141 double the 2.67% achieved in Australia. More significantly the rate of family following among Constituency MPs in New Zealand, who as explained above are the candidates who are incentivised to create and maintain a personal political machine, is even higher, at

8.45%. These findings support the Theory of Dynasty Formation but are exactly the opposite of what should be expected if openness rather than intra-party competition were the key factor in driving higher rates of family following.

In summary, none of the Alternative Hypotheses advanced can account for the variation in the rates of family following observed in the data and thus none constitutes a viable alternative explanation to the Theory of Dynasty Formation proposed in this thesis.

4.6 Additional Findings:

4.6.1 Age at First Election

As expected family followers do indeed start their parliamentary careers at a younger age than non-followers. The average age at which MPs from this study who are members of political dynasties were first elected is 40 years and 2 months old. Non-followers have to wait on average another 3 years and 7 months before entering parliament. This suggests that family followers enjoy a modest, yet appreciable advantage in launching their political careers. This may be related to the lesser need for family followers to serve an apprenticeship in local or regional government as outlined under Observable Implication 6 above.

142

As seen in the country specific findings above, the age gap between followers and non- followers varies from country to country, but is not correlated in any way to the rate of family following. Australia, with its very small sample size shows by far the largest gap, though one that runs in the other direction. Australian family followers have waited an average of 15 years longer than non-followers to be elected to the Federal Parliament. This might be thought to reflect the relatively high level of same-constituency following in

Australia, whereby family members have to wait for their relatives to vacate office before their own national careers can begin, a waiting time usually spent gaining experience in local or regional office. The opposite effect however is seen, albeit to a smaller degree, in both

Belgium and Ireland where same-constituency following is even more common than in

Australia.

4.6.2 Gender:

Family following is highly gendered. Outside of the United States, political dynasties are heavily male dominated. Even countries that score well on the overall proportion of women in their parliaments have few women amongst their family followers. Given, as we have seen, the tendency for family followers to achieve promotion to ministerial office at much higher rates than non-followers, this has wider implications for gender equality.

It may be worth noting that the only country in this study where women are more prevalent among family followers than among non-followers, the United States, is also the country with the lowest participation rate for women in parliament. Women do not feature at all among the current parliamentary political dynasties of the Netherlands, Australia and New

Zealand, countries with much better records than the US on overall female participation.

143

As stated previously, Ireland prior to 2007 also had proportionately more women among family followers than non-followers, but this was at a time when Ireland’s overall rate of female participation was even lower than that now seen in the US, a problem recently ameliorated by the introduction of gender quotas at the candidate level in 2016.

Graph 23: Women as % of all MPs by Country

45% 40% 35% 30% 25% 20% Family Followers 15% Non-Followers 10% 5% 0%

This suggests that in systems that are less supportive of female participation, dynasties do offer an easier route for women in to parliament, or at least for those few women who are fortunate enough to be related to previous office holders. The relationship between the inheritance of political machines and female participation is thus unclear. As can be seen below, if the US figures are included, then greater intra-party competition correlates with lower levels of female participation in general, but much higher rates of female participation amongst family followers. If the US is excluded however, then female participation amongst

144 both family followers and non-followers increases in the presence of intra-party competition, though the lack of women in political dynasties remains striking.

Graph 24: Family Following by Gender & Intra-party Competition

40.00%

35.00%

30.00%

25.00% % Women Family Followers 20.00% % Women Non Followers 15.00%

10.00%

5.00%

0.00% No Intra Party Intra Party Intra Party Competition Competition Competition ex (Neth, Can, Aus, (NZ Const, USA, USA NZ List) Bel, Irl)

4.7 Multi-variate Analysis

The comparative cross-national element of this study looked at 1,502 parliamentarians, of whom 102 or 6.79% are family followers. Setting family following as the dependent variable these results were subjected to a multi-variate regression analysis as outlined below.

According to the Theory of Dynasty Formation the key independent variable is the degree of intra-party competition fostered by the electoral system. Closed List PR as used in the

Netherlands is theorised to provide the least degree of intra-party competition and has

145 been used as the reference category. MMP has been divided into its two components List and Constituency and each has been treated as a separate system. The other electoral systems are ranked in order of increasing degree of intra-party competition.

A number of other independent variables that were identified in the literature were tested, principally urbanism, gender, education, a country’s CPI and HDI scores, and whether the party the MP represents is traditionally successful and whether it is socially conservative.

The results are in Table 14 below:

Table 15: Multi-Variate Analysis

Coefficient Std error p Electoral System MMP List - New Zealand 0.489 1.24 0.693 SMP – Canada 0.137 0.836 0.87 AV - Australia 0.212 0.909 0.816 MMP Const – New Zealand 1.376 0.871 0.114 SMP (open primary) - USA 0.949 0.774 0.22 Open-List PR - Belgium 2.384 0.759 0.002 PR-STV - Ireland 2.482 0.79 0.002 Other Factors Urban 0.301 0.296 0.31 Rural 0.466 0.324 0.151 Female -0.701 0.286 0.014 Graduate 0.289 0.292 0.322 CPI Score 0.052 0.012 0.072 UNHDI Score -6.141 11.859 0.605 Social Conservative Party -0.149 0.236 0.528 Traditionally Successful Party 0.727 0.359 0.043 Constant -4.737 0.7961 0.00 N 1502 Pseudo R2 0.109

The effect of electoral systems that promote intra-party competition – MMP Constituency,

SMP (open primary), Open-List PR and PR-STV - on the rate of family following is clear, though only in the cases of Open-List PR and PR-STV are the findings statistically significant.

146

Of the other factors that are suggested in the literature as possible influencers on the rate of family following two, gender and representing a traditionally successful party, are statistically significant. Both have an appreciable effect. MPs from traditionally successful parties are in all cases more likely to be family followers while women are in all cases less likely to be, though neither factor is as important an influencer as the electoral system.

Beyond that it is possible to say that family followers are somewhat more likely to represent rural areas and to be graduates and are somewhat less likely to represent socially conservative parties, though the lower significance levels for these factors reduces the degree of confidence with which such statements can be asserted.

Using the results outlined in Table 14, it is now possible to calculate the probability that an

MP will be a family follower under each of the electoral systems in this study. These probabilities are listed in Table 15:

Table 16: Probability Table Electoral Systems- controlling for all other factors

Likelihood MP will be Family 95% Electoral System Follower Confidence Intervals

CLPR - Netherlands 2.20% -0.009 0.053 SMP - Canada 2.50% 0.001 0.041 MMP List - New Zealand 3.51% -0.034 0.104 Alternative Vote - Australia 2.69% 0.001 0.053 SMP (open primary) - USA 5.43% 0.034 0.075 MMP Constituency - New Zealand 8.01% 0.019 0.142 Open-List PR - Belgium 19.10% 0.112 0.269 PR-STV - Ireland 20.60% 0.133 0.279

147

As can be seen above, controlling for other factors, the chances that a member of parliament will be a family follower is almost nine times higher under the PR-STV system used in Ireland and the Open-List PR system used in Belgium than under the Closed-List PR system used in the Netherlands. Constituency MPs in New Zealand are more than three and a half times more likely to be family followers than Dutch MPs, while members of the US

Congress elected under the open-primaries version of SMP are more than twice as likely to be the inheritors of a political dynasty than their colleagues in Canada.

As can be seen in Graph 25 below, the 95% Confidence Intervals for some of these findings overlap, making definitive statements impossible. Once other factors have been controlled for there is little difference in the probability that an MP will be a family follower under SMP in Canada, AV in Australia, CLPR in the Netherlands or SMP (open primaries) in the United

States.

Curiously the probability amongst list MPs in New Zealand is a little higher, though the significance levels of the findings for all five countries are above 0.05 and must therefore be treated with a degree of caution. For Canada, Australia and the US we know that the findings from this study are in line with those found in the work of other scholars (Godwin

2013; Lumb 2012; Lumb and Goddett 2015; Laband and Lentz 1985; Dal Bó et al. 2009).

148

Graph 25: Likelihood of Family Following by electoral system controlling for other factors at 95% Confidence Intervals 3 . g n i w 2 o . l l o f

y l i m 1 . a f

f o

d o o 0 h i l k L i 1 - .

t y s t R is P V c e is V P l M A n i l T L e r n S C P S u a e M it im p R M t r P s p O n n o e c p P o M P M M S

When other factors are controlled for the likelihood that MPs in Australia will be family followers is slightly higher than the likelihood that MPs in Canada will be. This reverses the positions produced by the headline data, but actually accords better with the predicted outcome of the Theory of Dynasty Formation as the slightly greater tendency for former party members to run as Independents in Australia should lead, in some cases, to a sort of quasi intra-party competition and thus to a marginally higher rate of family following. For the Netherlands and New Zealand additional research is needed to see if these findings remain consistent over time.

149

When other factors are controlled for the gap between the rate of family following in

Ireland and that in Belgium narrows quite significantly, though Ireland remains the most dynastic country in this study. This narrowing of the gap suggests strongly that it is intra- party competition that is the key driver of family following and that the ability of candidates to seek preference votes from other parties in Ireland is, as in Australia, a minor factor at best. The very clear gap between Ireland and Belgium and all other countries in the study strongly supports the Theory of Dynasty Formation.

4.7 Conclusions:

While exhibiting quite a degree of variability at the national level, the country level findings, bi-variate analysis and multivariate analysis all support the central hypotheses outlined in the Theory of Dynasty Formation that the the degree to which the electoral system incentivises the creation and maintenance of a personal political machine is the strongest explanatory variable for the observed differences in the rate of family following. None of the alternative hypotheses are similarly supported.

The next chapter will look at family following in a single country, Ireland, to test these hypotheses over a prolonged period of time. Ireland is being looked at as, among the countries in this study, it has both the highest rate of family following in relative terms and the highest number of family followers in absolute terms. Ireland thus provides a larger number of cases to establish more credible findings and constitutes what Seawright and

Gerring (2008) describe as an “extreme case”. They argue that there is considerable value in

150 examining such an extreme case as part of a broader comparative study (Seawright and

Gerring 2008).

Furthermore, in the Nealon’s Guides, Ireland also has an easily accessible source of detailed biographical information on all members of parliament going back several decades. Finally,

Ireland was indentified over 40 years ago as one the “traditional societies” that clung to family following as a result of underdevelopment (Putnam 1976). Since then Ireland has developed rapidly and now boasts the 4th highest GNI per capita in the EU. If there is any residual merit to the Putnam thesis, we should find it in a declining rate of Irish family following over the last four decades.

151

Chapter 5: Longitudinal Case Study: Ireland

5.1 Three funerals:

In the space of less than thirty days in late May and early June of 2011, the deaths occurred of three of the most influential politicians in recent Irish history – former Taoiseach Garret

Fitzgerald, former Attorney General and former Minister for Finance Brian

Lenihan. All three were the sons of former leading politicians. Garret Fitzgerald's father

Desmond was Minister for External Affairs from 1922 to 1927, a role his son was to take on as Minister for Foreign Affairs five decades later (Nealon 1981: 70). Declan Costello had a similar experience. When appointed Attorney General in 1973, he took over a role his father, the former Taoiseach John A Costello, had relinquished in 1932 (Fanning 2011).

Brian Lenihan too followed exactly in his father's footsteps. When first appointed to cabinet in 2007, it was as Minister for Justice (Collins 2007: 94), the same portfolio that his father, the former Tánaiste Brian Lenihan Snr held as his own first ministerial appointment from

1964 to 1968.

That senior ministers should be the sons of former politicians is not unusual in Irish politics.

Seventeen men have served as head of government in Ireland. The first seven: Cathal

Brugha, , Michael Collins, WT Cosgrave, Eamon DeValera, and

John A Costello (Taoiseach 2013) were members of the revolutionary generation that set up the state and therefore could not have inherited political power. Of the ten Taoisigh since

1966 (Taoiseach 2013), four, Liam Cosgrave Garret Fitzgerald (Nealon 1981: 70), Brian

Cowen and Enda Kenny, have been the sons of former TDs, while one, was

152 the son-in-law of a former Taoiseach. In the 30th Dáil, from 2007 to 2011, the Taoiseach,

Tánaiste, Minister for Finance and Leader of the Opposition were all the children of TDs. All four represented the same constituencies as their fathers. Three of the four were first elected in by-elections immediately following their fathers' deaths. In the fourth case, that of Mary Coughlan, there was no by-election held after her father died in a car crash as a general election was due within months. She duly won that election and the seat that had previously been held by both her father and her uncle.

5.2 Outline of Case Study:

This chapter examines the phenomenon of inherited political power in a single country, the

Republic of Ireland over a period of eleven elections from 1981 to 2016. The election results are analysed using the same methodology outlined in Chapter 3. The biographical details for the TDs are sourced, as with the main study, from Nealon's Guides a work published after each election providing detailed information on each member of the incoming lower and upper houses of the Irish parliament, the Dáil and Seanad. Taking this approach allows the hypotheses that underpin the Theory of Dynasty Formation to be tested over a significant period of time thus reducing the risk that an anomolous event could skew the findings.

1,818 seats were filled in these 11 elections. Of these, 423 or 23.37% were won by family followers. 594 different candidates were elected to fill those seats. Of these 125, or 21.04%

153 were members of political dynasties. This longitudinal study therefore affords a much larger sample size and should produce results that provide greater confidence.

These eleven elections cover a period of remarkable economic and social change in Ireland.

In 1981 Ireland was one of the poorest countries in western Europe. By 2008 it was, in terms of GDP per capita, the 2nd richest. From stagnation in the 1980’s through the boom years of the “”, the economic collapse of 2008 and the subsequent recovery,

Ireland has endured significant economic dislocation, including the decline and closure of traditional industries, a flight from the land to the cities and the development of a modern knowledge based economy. The social changes over these years were just as dramatic incorporating a dramatic decline in the status of the Catholic Church, increasing liberalisation of social mores and values, a radical expansion of 3rd level education and, since 2002, mass immigration for the first time in modern Irish history. Politically this period saw the ending of the conflict in , the emergence and disappearance as Dáil parties of the Workers Party, Democratic Left and the and the entrance into the Dáil chamber of the Green Party and Sinn Féin. Both of Ireland’s main political parties experienced an electoral collapse during the period: Fine Gael in 2002, and

Fianna Fáil in 2011. The smaller Labour party, which tends to run just a single candidate in each constituency and thus is less exposed to intra-party competition, also suffered a severe electoral backlash in 1997 and 2016.

154

One thing that has not changed over this period is Ireland’s electoral system or the ferocity of the intra-party competition it fosters. If the Theory of Dynasty Formation outlined in

Chapter 2 is correct, then despite the dramatic changes that Ireland has undergone over the

35 years from 1981 to 2016, the rate of family following in the Irish Dáil should remain high throughout. It is expected however that when either of the two main parties, who traditionally run multiple candidates in most constituencies, suffer an electoral collapse that the rate of family following in those years will decline. An electoral collapse by a party which tends not to run multiple candidates, even a severe collapse such as that suffered by the Labour party in 2016 by contrast, should not lead to a decline in the number of family followers in the Dáil.

5.3 Extended Hypotheses:

To determine whether this is in fact the case, this longitudinal study will test the results of the 11 elections between 1981 and 2016 against the relevant hypotheses, alternative hypotheses and observable implications from Chapter 3.

Hyptheses 1 to 4 can be reduced to a single question: is the rate of family following in

Ireland, which combines full intra-party competition with preference vote seeking from other parties, consistently higher throughout the period than that seen in other countries?

We know from existing studies that the long term rate of family following in Canada is 3%

(Godwin 2013), in Australia it is 4% (Lumb 2012; Lumb and Goddett 2015) and in the United

155

States it is 6% (Dal Bó et al. 2009). Fiva and Smith also found a consistent rate of family following in Norway of 7% (Fiva and Smith 2016).

All four of these studies use a methodology that results in a higher figure for family following than that used in this study; including relatives beyond two degrees of separation and, where the founder of a dynasty and a successor serve concurrently, counting both rather than just the follower as has been done here.

To support Hypotheses 1 to 4 family following in Ireland would have to be, consistently over the period, of a magnitude significantly greater than that seen in the longitudinal studies mentioned above in Canada, Australia, the United States and Norway and should be consistently higher than the rates of family following found in New Zealand and Belgium in the comparative cross-national section of this study.

Alternative Hypotheses, 1 and 2, that higher rates of family following correlate with lower relative levels of development at a national level or with higher levels of perceived corruption at a national level cannot be tested on a sub-national basis and are omitted from this case study.

The question of local control of the nominating process will not be looked at. As outlined in

Chapter 2, despite the fact that most Irish political parties have a theoretically de-

156 centralised system of candidate selection, in reality party headquarters have a very significant degree of control over the process, making designations such as “locally controlled” somewhat meaningless. Furthermore, when Smith (2012) tested the question across forty nine parties in eight countries, including Ireland, he did not find support for the hypothesis outside of Iceland.

The other Alternative Hypotheses; that the rates of family following is related to the proportion of TDs who represent rural areas, who are graduates or who are members of socially conservative parties are testable and will be tested against the results of the eleven elections in the study.

Similarly, Additional Observable Implication 5 relates to the MMP electoral system and cannot be tested on a sub-national basis in Ireland. Four other Additional Observable

Implications are testable and will be checked against the results of all eleven elections. The modified observable implications are:

 That instances of family following will tend to be in the same constituency as

the preceding family member.

 That instances of same-constituency family following will tend to occur

directly following the death, defeat or retirement of the preceding family

member before their personal political machine has time to decay.

 That family following will be more prevalent among more traditionally

successful parties, as such parties are likely to evidence higher rates of intra-

157

party competition. This should also hold true for former members of such

parties who choose to run as Independents.

 That family followers will be disproportionately represented among high

office holders such as cabinet ministers.

Additional Observable Implication 6 is technically a cross-national measure, but the Theory of Dynasty Formation strongly suggests both that :

 Most TDs will begin their careers in local government and

 That in the years following the ending of the dual mandate in 2003 that

the number of Family Followers with prior experience as councillors

should rise sharply. This proposition too will be tested.

5.4 Why Ireland?

As discussed in Chapter 4, Ireland shows the highest rate of family following amongst the countries in this study. This is as expected given that Ireland’s electoral system combines full intra-party competition with preference vote seeking from other parties and therefore incentivises to the greatest extent possible, among the countries in this study, the creation and maintenance of personal political machines.

To test the hypotheses further, and to control for transitory explanations, this chapter looks, not at a single election, but at a series of elections over a number of decades. It also examines alternative explanations that have been offered for Ireland’s higher rate of family following.

158

As we have seen in Chapter 2, over 50 years ago Robert Putnam (1976) identified a small group of “semi-traditional societies” namely “Morocco, Sri Lanka, Lebanon, Ireland, Greece and Republican China” where family following remained a significant factor in elected politics. Putnam did not define what a semi-traditional society might be and his list of countries is somewhat eclectic, eschewing for example India (Besley 2005: Page 51)

Colombia (Acemoglu and Robinson 2006: 26), Mexico (Camp 1982: 850-851) and the

Philippines (Riedinger 1995: 209), where political dynasties have long dominated democratic politics. The phrase is used in a dismissive manner to suggest that the countries listed are somehow outside the democratic mainstream.

Putnam is not the only commentator to have taken the view that Irish democracy or at least democracy in certain parts of Ireland is less than fully developed. Mair (1981) drew a clear distinction between loyalist voters in the west of Ireland – a geographical entity he defined as Connaught plus Clare and Kerry – who tended to vote for the same party in each election and instrumentalist voters in the east of the country who voted on the issues. According to

Mair (1981), loyalist voters were more likely to come from rural areas, to be poorer, less well educated and to inherit their political outlook. Carty (1980), in his study of politics in the rural eastern county of Kildare referred to Ireland's “peasant political culture”. Saks

(1976), in his book on rural politics in Donegal on the north-west coast, did acknowledge that Ireland was not a classic clientelist regime where politicians traded state resources for votes, but suggested that to a degree this didn't matter as the politicians fooled voters into thinking that it was. All three accounts describe politics, in rural Ireland at the beginning of

159 our period of study at least as being dominated by traditional, clientelist values. Komito

(1984) took a different view. What Saks saw as clientelism and thus evidence of state capture, he identified as brokerage, an attempt by political candidates to differentiate themselves from one another by the level of personalist constituency service they offer, in effect by establishing personal political machines.

This thesis sees the incidence of family seats in Ireland, and elsewhere, as being directly linked to Komito's (1984) concept of brokerage. While some level of brokerage occurs in all political systems, it is more prevalent in Ireland because the electoral system involves the use of multi-seat constituencies and therefore requires politicians to compete against fellow party members. 56% of the Fianna Fáil TDs who lost a seat between 1927 and 1997 did so to a party colleague (Gallagher 2000). Such intra-party competition results in both politicians and voters putting more emphasis on brokerage. As outlined in Chapter 2, the need to provide a personalist constituency service independent of one's running mate requires the politician to build and maintain a personal political machine, separate from the electoral machinery of their political party, capable of servicing the brokerage demands of their constituents. Because the machine belongs to the politician and not the party, the politician has a significant say in who inherits it.

5.5 Inheriting a Machine

In order to be elected in Ireland, a candidate requires a significant support network. If he or she wishes to run on behalf of a party, they must first win that party's nomination. For all

160 major Irish political parties this usually means being chosen by a selection convention where the vote is restricted to existing party members of several years standing. Once selected the candidate may have to compete not just with the candidates of other parties, but with one or more candidates from their own party. The candidate, while receiving some support from their part, will generally require money to pay for posters, leaflets and advertising; campaign workers to put up the posters, distribute the leaflets, knock on doors and ask for votes and they will require an office and a volunteer staff to run it. Most of all they will require contacts; with the media, with community organisations and sports clubs, with pressure groups and lobbies and with key decision makers in local and national government.

Studies show that a significant minority of voters vote on a candidate rather than a party basis (Bowler and Farrell 1991; Marsh 2007), but this only tells half the story. A TD must not only pitch for these floating personalist voters, they must appeal to the voters who support their own party to choose them ahead of their running mate (Gallagher 2008). Such a case can best be made on a non-party basis by appealing to localism, factionalism or personalist brokerage.

The central contention of this thesis is that to be successful in a multi-seat constituency running against members of their own party, a candidate cannot rely on an ideological or policy based appeal to get elected. Such a platform would be shared with their opponents from within their own party and cannot help them overcome those opponents. Instead they must develop a machine of their own that will stand in a semi-independent relationship to their political party.

161

Most TDs build their machines over time. They spend several years in local government or community organisations, developing contacts and growing a support network. They issue press releases, make speeches at party meetings, post on social media and seek to be identified with key issues. They sign up relatives, friends and neighbours as delegate votes at convention and bide their time waiting for their opportunity to contest for the Dáil.

Others simply inherit the lot – the delegate votes at convention, contacts in party headquarters, name recognition with the public, the network of clients, fund-raisers, canvassers and election workers whose primary loyalty is not to the party but to their close family relation, the previous Oireachtas member. The advantage this gives the candidate from the political family over an opponent from the same party who lacks a dynastic background is considerable, particularly in their first election, and may be the most important element explaining the prevalence of family seats and the wider success enjoyed by their holders.

5.6 How is a Political Machine inherited in Ireland?

As outlined in Chapter 3, the inheritance of a personal political machine can be through one of three main routes:

 Direct inheritance. By far the simplest way of transferring a machine from one

candidate to another is when a vacancy occurs through the death, defeat or

retirement of a TD. As long as there is a clearly recognised successor the machine

should pass from one candidate to another effectively intact. The ending of the dual

mandate in Ireland in 2003, when Oireachtas members were barred from holding

162

seats on local authorities has, if anything, made direct inheritance more

straightforward. Previously it was normal practice for Oireachtas members to serve

as councillors in their own local electoral areas, both to keep an eye on local affairs

and to prevent a party rival undermining their electoral base. The ending of the dual

mandate posed a difficulty. In handing their council seat over to someone else,

would the Oireachtas member be facilitating a future competitor? The solution for

many Oireachtas members was to appoint a close family member to their council

seat. This not only protected the Oireachtas member's base from intra-party

competition, it also sent a clear signal to his or her machine, and to the general

public, who their chosen successor might be.

 Delayed inheritance. Such smooth successions are not always possible. The death of

a TD or their defeat, often by a member of the same party, may happen

unexpectedly. In such circumstances it may be a considerable period of time before

a family member is in a position to run to regain the seat. In the interregnum the

personalist brokerage relationships built up by the previous Oireachtas member may

weaken or be usurped by an opponent; control of the local party organisation may

be lost. The longer the gap between the original Oireachtas member losing their

seat and their family member running for election the less of the machine is

available to be passed on.

 Extra-constituency inheritance. Most political dynasties consist of a child following a

163

parent in the same constituency, but not all. Certain families – the De Valera1,

Andrews2, Lenihan3, Lemass4, and Kitt5 families in Fianna Fail, the Collins6, Bruton7

and Belton8 families in Fine Gael and the Lynch9 family in Labour have all had

multiple members in Dáil Éireann often at the same time and representing different

constituencies. In such a case one cannot speak of such candidates inheriting a

constituency machine. They can however inherit certain aspects of the machine –

name recognition and contacts in party headquarters being the most obvious

elements. Less tangible, but no less important is the on the job training being part of

a political family and participating in elections from an early age can provide

(O’Rourke 2012).

In Ireland there is a fourth route:

1 Eamon DeValera, founder of Fianna Fáil and Ireland's longest serving Taoiseach represented the Dáil constituency of Clare. He was followed into Dáil Éireann by his son Vivion, who represented North West, grand-daughter Silé De Valera, who represented Clare and grandson Éamon O'Cuív who represents Galway West (Nealon 1992: 98; Nealon 1987: 20). 2 David Andrews represented the Dáil constituency of Dún Laoghaire. He was followed into Dáil Éireann by his brother Niall who represented Dublin South, his son Barry who succeeded him in Dún Laoghaire and his nephew Chris who represented Dublin South East. (Collins 2007: 86) 3 Brian Lenihan who represented Roscommon and later Dublin West in Dáil Éireann, was followed into the Dáil by his father Patrick who represented Longford-Westmeath, his sister Mary O'Rourke who succeeded their father in Longford-Westmeath, his son Brian who succeeded him in Dublin West and his son Conor who represented Dublin South West (Collins 2007: 94). 4 Seán Lemass former Taoiseach who represented Dublin South was followed into Dáil Éireann by his son Noel who represented Dublin South West, his daughter in law Eileen who succeeded Noel in Dublin South West, his son-in-law Charles Haughey another Taoiseach who represented Dublin North East and his grandson Seán Haughey who represented Dublin North Central (Nealon 1981: 50; Nealon 1992: 59) 5 Michael F Kitt who represented Galway East was followed into the Dáil by his son Michael P Kitt who also represented Galway East, by a second son Tom Kitt who represented Dublin South and by his daughter Áine Brady who represented Kildare North (Collins 2007: 102) 6 General Michael Collins was Minister for Finance in the first Dáil and Ireland’s shortest serving head of state, holding the position of President of the Council for just ten days. He represented the constituency of Cork Mid, North, South, South East and West. He was followed into Dáil Éireann by his sister Margaret Collins-O'Driscoll representing Dublin North and by his grand-niece also representing Dublin North (Nealon 1981: 47; Sheridan: 2010). 7 former Taoiseach who represented Meath was followed into Dáil Éireann by his brother Richard who represents Dublin North Central (Nealon 1982: 49) 8 who represented Dublin North was followed into Dáil Éireann by his sons Jack and Paddy Belton both of whom represented Dublin North East, his nephew Luke who represented Dublin North Central and his granddaughter who represented (Nealon 1983) 9 Kathleen Lynch who represents Cork North Central was followed into the Dáil by her brother-in-law Ciarán Lynch who represents Cork South Central (Collins 2007:47)

164

 Inheritance through the Senate. Having a Senator in the family may not be quite as

useful to an aspiring TD as having a Dáil member. Senators are not directly elected

by the public and their public profile is usually lower than that of a TD, but it can

convey similar advantages. Many Senators are aspirant or recently defeated TDs and

use their time in the Upper House to build, repair or nurture their political machines.

While officially Senators have sectoral rather than geographical constituencies, in

reality many of those intending on standing for the Dáil act as additional members

for their future Dáil constituency.

What about Independents?

Thus far we have spoken solely of members of political parties, but PR-STV also allows the election of large numbers of independent TDs. Ireland is the only country in the EU (and the only country in this study) that regularly elects large numbers of Independents to the lower house of it’s parliament (Ehin et al. 2013). Though not needing delegate votes to be selected to run, independent candidates still require their own political machines to achieve electoral success. This machine is regularly passed on to a family member. Examples include the Blaney family in Donegal North East, where Harry Blaney succeeded his brother

Neil and was in turn succeeded by his son Niall; the Fox family in Wicklow, where Mildred

Fox was elected to the seat previously held by her father Johnny and her brother

Christopher took over his council seat; the Healy-Rae family in Kerry South, where Michael

Healy-Rae won the seat vacated by his father Jackie, and was recently joined in the Dáil by his brother Danny with both TDs being succeeded in their council seats by Danny’s children

Johnny and Maura Healy-Rae; or the emergent Lowry dynasty in Tipperary North where

165

Micheál Lowry, son of TD is a county councillor, leads a group of six independent councillors elected under the banner of “the Lowry team” and is widely tipped to take over the Lowry machine upon his father's retirement.

Why, given that they are not competing with fellow party members, do independent TDs and their supporters so readily adopt a clan-based approach to political succession? A possible explanation is that regardless of what issue first propels them into office, to survive under Ireland’s electoral system most independent TDs must provide a personalist, brokerage based service. As such they constantly compete with every other localist politician in the field and have to develop a machine capable of matching those of their opponents. Where Independents lack the unifying appeal of either a party or an ideology, they fall back on family as the social glue to hold that machine together. Once built, such a machine is largely indistinguishable from those of party politicians and can be passed on in the same way. It would follow that those Independents who are elected primarily for localist reasons such as hospital campaigns are more likely to develop into political dynasties.

Such an argument for family following among Independents should of course hold in other countries with personalised voting. However the sheer lack of Independents in the lower houses of other mature, developed, democratic states makes any effective comparison difficult. Of the countries in this study only one other, Australia, has any Independent members of its lower house at all. 25% of Independent Australian MPs are family followers,

166 but as this figure represents just one man – Bob Katter MP– it can hardly be regarded as statistically significant.

In Ireland a perhaps more important factor is that many Independent TDs emerged from the political parties. For example the founding members of the Blaney, Fox and Healy-Rae dynasties, and the nascent Lowry one, began their careers as members of political parties.

They chose to run as Independents for various reasons; failure to secure a Dáil nomination through the party in the case of the Healy-Raes and Foxes, policy differences over Northern

Ireland in the case of the Blaneys and expulsion from the party in the case of Michael Lowry.

These so called “gene pool” Independents often compete on similar policy platforms as their erstwhile party colleagues and some, such as Beverley Cooper-Flynn in Mayo, are eventually re-absorbed into their former parties. Such Independents act, in effect, as quasi party candidates, increasing the level of intra-party competition, and the incentive for personal machine building, still further. It is perhaps noticeable that three of the five Independent

TDs in the current Dáil who are members of political dynasties, , and Michael Healy-Rae, are former members of political parties, and that the other two

Independent family followers, Danny Healy-Rae in Kerry and Seán Canney in Galway East, followed relatives who were themselves former members of political parties. Indeed in every election from 1981 to 2016, not only have most Independent TDs been former members of political parties, every single Independent TD who is a member of a political dynasty has either been a prominent party official or an elected representative for a political party or has inherited their seat from a relative who was.

167

Smith (2012) observed a similar process in Japan under their former Single Non Transferable

Vote electoral system. Candidates who failed to secure a nomination from the Liberal

Democratic Party (LDP) would sometimes run as Independents targetting that party’s voters. If successful in winning a seat in the Diet they would promptly rejoin the party leading to the aphorism “if you win, you’re LDP”. To successfully pull off such a stratagem a candidate needed a powerful personal political machine. Unsurprisingly Smith (2012) found that it was members of political dynasties who resorted to this tactic most frequently. Again

Australia provides support for this idea of at least some Independents as quasi party candidates. All four of Australia’s Independents are former members of other political parties; two National, one Liberal and one Green. All four competed against their former parties, in effect providing, unusually for Australian politics, a degree of intra-party competition in those constituencies.

5.7 Methodology:

The source for the biographical information on all TDs elected in the 11 elections between

1981 and 2016 are the relevant Nealon's Guides (Nealon 1981, 1982, 1983, 1987, 1989,

1992, 1997; Kennedy 2002; Collins 2007, 2011, 2016). These guides have long been seen as definitive and as superior in their accuracy to similar publications in other countries (Farrell

1970). Relying solely on the guides ensures a consistency of approach.

Consistency and ease of comparison are also the reason for beginning this study in 1981.

That was the year the Dáil reached the size of 166 members that it retained for all elections

168 until 2016 when the size was reduced to 158. This was also the first election where the constituency boundaries were set by an independent commission (Mair 1981). There has been only one small change in the size of the national parliament over the period under study, a reduction of eight seats in 2016, and only limited change in individual constituencies. This consistency makes geographical and temporal comparisons across elections easier and more reliable. A study of family seats beginning with the 1st Dáil in

1919 would serve little purpose because, as Farrell (1970) has pointed out, family following only became a major route of entry in Dáil Éireann as the revolutionary generation began to retire or die out in the 1960s. The eleven elections being looked at in this chapter give a total of 1,818 individual elections of 594 different candidates to Dáil Eireann, a sufficiently large number for conclusions to be drawn from the evidence with a reasonable degree of confidence. By-elections in Ireland differ in that a single candidate is run for each party.

Though, not officially acknowledged as such, they are thus, in effect, conducted using the

Alternative Vote system and have therefore not been looked at.

Each Nealon's Guide contains the results of the election broken down by constituency, a brief biography of each TD elected including any current government or parliamentary roles that they hold, their previous employment, educational and electoral history and, crucially from the point of view of this study their familial relationship to previous Oireachtas members. The guides have often included a summary of the number of family seats in their statistics section, but these summaries have not been relied upon for this study. Some editions – 1981 and 1982 – did not include such a section on family seats at all, while the others have only included TDs who followed a relative who was a member of the Dáil and

169 ignored the possibility of inheritance through the Seanad; the figures for family seats given in the statistics section of these editions are broken down by gender, but not by constituency, party, membership of the cabinet or prior education. For these reasons, this case study relies instead on the raw data contained in the individual biographies. The TDs returned at each election have been listed in constituency order and cross tabulated for the various factors to be measured, including whether they represent a traditionally successful or socially conservative party, whether they are a graduate, any family relationship with a previous Oireachtas member, membership of the cabinet, gender, previous experience as a local councillor, and method of inheritance; direct, delayed, extra-constituency or through the Seanad. As with the comparative cross-national results detailed in Chapter 4, a simple binary measure has been employed, 1 if the factor is present and 0 if it is not. For gender 1 is used for women and 0 for men.

5.8 Results:

5.8.1 Extended Hypotheses 1 to 4: That the rate of family following in Ireland is consistently higher than that found in Canada ( 3%), in Australia (4%), in the United States

(6%) or in Norway (7%) and higher than the rate of family following found in New Zealand

(6%) and Belgium (15%) in the cross-national portion of this study

The results, as can clearly be seen in the graph below, strongly support the extended hypotheses. The nine general elections between 1981 and 2007 show a consistently high level of family following in Dáil Éireann, always in excess of 22.8%. In only one of these nine

170 elections over twenty six years did the proportion of family seats vary outside a narrow band of just two and half percentage points. That was in 2002 when the electoral collapse of the Fine Gael party saw the rate of family following fall to 22.89%, 3.01% of the peak rate of 25.9% recorded in 1997. Even this decline was completely reversed at the following election in 2007 when the rate of family following rebounded to 25.3%.

The 2011 election did see a dramatic fall in the number of family followers to 15.7%, but even this is at a rate still higher than that seen in any other country in this study including

Belgium. The decline of 2011 is entirely accounted for by the loss of family seats in the, until then, dominant Fianna Fail party.

The 2011 election was a shock to the Irish party system. Fianna Fail, which had won the most seats in every general election since 1932 plummeted to third place, losing 51 of their

71 Dáil seats including 20 of their 26 family seats. The number of family followers in the newly dominant Fine Gael party actually rose at this election. Fianna Fail’s partial electoral recovery in 2016, when they won 44 seats including 12 family seats has seen the overall proportion of family seats rise once more to almost 19%. This combined with the consistent finding in opinion polls from June 2017 to April 2018 of a surge in support for both main parties (Kavanagh 2018) strongly suggests that the 2011 dip was an anomaly.

171

Graph 26: Rate of Family Following Ireland 1981 - 2016 30.00%

25.00%

20.00%

15.00% Family Seats

10.00%

5.00%

0.00% 1981 1982 1982 1987 1989 1992 1997 2002 2007 2011 2016 Feb Nov

As posited in the Theory of Dynasty Formation, dynasties take time to form. A shock to a party system, such as happened in Ireland in 2011, should lead to a sudden, if temporary, fall in the rate of family following, before the inherent nature of the system causes the rate to rise again. Smith (2012) observed a similar temporary decline in the rate of family following in Japan after the economic crash of 1992 saw the ruling LDP suffer heavy, but temporary, electoral losses.

5.8.2 Additional Observable Implications:

Five of the Additional Observable Implications tested cross-nationally in Chapter 4, can be tested sub-nationally in Ireland over the eleven elections since 1981.

172

Additional Observable Implication 1: That most instances of family following will be in the same constituency.

As seen below, family TDs are consistently more likely to represent the same constituency as the family member they have followed into the Dáil and although the gap between the two narrowed in 1992, it has since widened again. On average over the period in question, a family TD was two and a half times more likely to represent the same constituency as their relative than they were to be elected elsewhere.

Given the emphasis placed on machines in this study, the only surprise, perhaps is that the gap is not even wider. One possible explanation for this is that, despite having multi-seat constituencies, it is incredibly rare for members of the same family to represent the same constituency at the same time. In fact it has occurred only twice in the period covered by this study, when Myra Barry won the 1979 Cork North East by-election and briefly served alongside her father prior to his retirement at the next election, and in 2016 when Danny

Healy-Rae was elected on his brother Michael’s surplus votes in Kerry. The reluctance of political dynasties to run multiple candidates in the same constituency, presumably for fear of splitting their vote and putting their existing seat at risk, means that politically ambitious brothers and sisters of TDs, who would have been subject to the same socialisation forces, usually have to carve a niche for themselves elsewhere. Examples of this would include

Bertie Ahern's brother Noel, Brian Lenihan's brother Conor, John Bruton's brother Richard,

Jim Mitchell's brother Gay and Michaels Kitt’s brother Tom and sister Áine. There were either 41, 42 or 43 Dáil constituencies during the period under study. That the clear

173 majority of family followers did not seek their fortune in one of the other ones, but patiently waited for a vacancy in their family's home patch is strong evidence in support of the hypothesis. Observable Implication 1 is held to be supported throughout the period under study.

Graph 27: Same Constituency 1981 - 2016

85.0%

75.0%

65.0%

55.0%

Same 45.0% Constituency Percentage

35.0%

Different 25.0% Constituency

15.0%

5.0% 1981 1982 – 1982 – 1987 1989 1992 1997 2002 2007 2011 2016 Feb Nov Election

Additional Observable Implication 2: That most instances of family following will occur directly following the death, defeat or retirement of the preceding family member before their personal political machine has time to decay.

As shown below, it is clearly the case that most succession is direct and follows immediately on the death, defeat or retirement of the preceding family member. Given that,

174 statistically, delayed inheritance should be more likely, both because more family members would reach adulthood and be available to run for election and because more elections give more chances of being elected, this is very strong evidence in support of the contention that what is being inherited is the former TD’s political machine.

Graph 28: Direct v Delayed Inheritance 95.0%

85.0%

75.0%

65.0%

55.0% Direct 45.0% Inheritance

35.0% Delayed Inheritance 25.0%

15.0%

Percentage of Family Followers Family of Percentage 5.0% 1981 1982 – 1982 – 1987 1989 1992 1997 2002 2007 2011 2016 Feb Nov General Election

Observable Implication 2 is held to be supported throughout the period under study.

Additional Observable Implication 3: That family following will be more prevalent among more traditionally successful parties, as such parties are likely to evidence higher rates of intra-party competition. This should also hold true for former members of such parties who choose to run as independents

175

Graph 29: TDs from Traditionally Successful Parties 100.00% 90.00% 80.00% 70.00% 60.00% 50.00% Traditionally successful parties 40.00% Other parties & Independents 30.00% 20.00% 10.00% 0.00%

In the early elections in this study almost all the TDs in Dáil Éireann, were members of one of Ireland’s three traditionally successful political parties - Fianna Fáil, Fine Gael and Labour

– or were “gene pool” Independents, such as the Blaney dynasty in Donegal, with a strong connection to one of the main parties.

Irish politics became somewhat more diverse in 1987, in party terms, with the entry into the

Dáil of the Progressive Democrats and then steadily less so over the four elections that followed as that party’s support declined. Diversity returned in 2002 when an electoral collapse for Fine Gael coincided with sharp rise in the number of Independents. Main party dominance resumed in 2007 and it is only in the last two elections that the proportion of seats held by traditionally successful parties has significantly fallen.

176

According to the Theory of Dynasty Formation, family following should be concentrated among traditionally successful parties. The candidates standing for such parties are far more likely to have running mates and therefore have a much stronger incentive to create and maintain a personal political machine. Smaller parties rarely run more than one candidate per constituency and therefore endure much lower levels on intra-party competition. Traditionally successful parties have also been around for much longer.

Dynasties take time to form and a party new to the Dáil, such as Sinn Féin, have yet to see a second generation of TDs succeed their first. As noted above however, in the early elections in this study there were almost no TDs in Dáil Éireann who were not members of traditionally successful parties. It is thus really only since 2011, when the proportion of TDs from non-traditional parties rose sharply, but the proportion of family followers from non- traditional parties did not, that the expected observable implication becomes irrefutable.

Graph 30: % of TDs who are Family Followers

30.00%

25.00%

20.00% Traditionally Successful Parties

15.00% Other parties & Independents

10.00%

5.00%

0.00% 1981 1982 1982 1987 1989 1992 1997 2002 2007 2011 2016 Feb Nov

177

There is a remarkable similarity between the shape of the darker line in Graph 26 which depicts the rate of family following in the whole Dáil and that in Graph 30 which depicts just the rate of family following in traditionally successful parties. It will be noted however that there was a notable spike in family followers from non-traditional parties in 1987. This coincides, as mentioned above, with the birth of the Progressive Democrats, a party founded by a member of a Fianna Fáil political dynasty: Desmond O’Malley. The

Progressive Democrats (PD), like Fianna Fáil and Fine Gael were a centre-right political party.

Not only did they compete for votes directly from these parties, they also recruited a number of dynastic standard bearers from both Fianna Fáil and Fine Gael to stand for them, thereby circumventing the usual time delay for a new party in establishing its first dynasties.

All four PD family followers elected in 1987 were members of existing Fianna Fáil or Fine

Gael dynasties and all subsequent PD family followers were members of the same ongoing dynasty, that of party founder, Desmond O’Malley.

The consistency of the rate of family following among the traditionally successful parties is remarkable with the only significant decline being in 2011. But while family following in the

Dáil has yet to return fully to it’s pre-2011 levels, the rate of family following in the major political parties was completely restored in 2016. Additional observable implication three is clearly supported throughout the period of this study.

Additional Observable Implication 4: That family followers will be disproportionately represented among high office holders such as cabinet ministers.

178

This implication has been tested against the membership of the eleven cabinets formed in the aftermath of the elections under study. Each involved the appointment of 15 members

(including Taoisigh) yielding some 165 cases. Re-shuffles have not been included, nor has the Rainbow cabinet of 1994-97 as these do not correspond directly with general elections, however their make-up was not significantly different from the other cabinets of the era.

As can be seen in the graph below, all eleven elections to 2016 led to a consistently higher level of representation of members of political families in cabinet than in the Dáil in general.

On no occasion throughout these more than 35 years were the holders of family seats under-represented at cabinet, including during their lowest ebb in the Fianna Fáil-Labour coalition of 1992-94, when they still made up 4 out of the 15 members of government or

26.7%. The cabinet formed after the 2016 election, by contrast, boasted no fewer than 7 family followers, making it at 47% the most dynastically dominated cabinet in modern Irish history. The evidence for the proposition that family followers will be more highly represented in cabinet is therefore supported.

179

Graph 31: Family seats in Dáil & Cabinet 1981 - 2016

50.0%

45.0%

40.0%

35.0%

30.0%

25.0% % of Family seats – Dáil 20.0% Percentage

15.0% % of Family seats – Cabinet

10.0%

5.0%

0.0% 1981 1982 – 1982 – 1987 1989 1992 1997 2002 2007 2011 2016 Feb Nov Election

Additional Observable Implication 5: Most TDs will begin their careers in local government and that in the years following the ending of the dual mandate in 2003 that the number of

Family Followers with prior experience as councillors should rise sharply.

As seen in Chapter 4 Irish TDs are more likely to have cut their teeth in local government than the members of any other parliament in this study. The graph below shows that, for family followers at least, this has not always been the case. While a clear majority of non- followers have always built their political base in local government before pursuing a national career, prior to the 2007 election, a high proportion of family followers skipped that stage, instead entering politics for the first time in a general election. Since 2007 the

180 gap in prior local government experience between followers and non-followers has all but disappeared.

Graph 32: Prior Experience in Local Government

85.00%

75.00%

65.00%

55.00% Family Followers Non Followers 45.00%

35.00%

25.00% 1981 1982 1982 1987 1989 1992 1997 2002 2007 2011 2016 Feb Nov

The key causal factor in this conversion of family followers to the merits of prior local government service was the abolition of the dual mandate in 2003. Before that date most

TDs also served as local councillors. Ministers were required to abdicate their council seats, but backbenchers were not. By holding a council seat, a TD could effectively prevent a local rival from building a base in their own territory. Because the Dáil seat took priority over the council seat, a succeeding family member would seek to inherit at the national level first and would acquire the local seat at a later date.

181

The abolition of the dual mandate reversed this order of succession. As sitting TDs were required to forego their council seats in advance of a general election, many sought to do so in a manner that would not empower a potential opponent and weaken their own chances of re-election to the Dáil. Vacated council seats in Ireland are not filled by by-elections, but by co-options. The situation is different for councillors who were elected as members of political parties than it is for those who are elected as Independents. If a councillor elected on a party ticket resigns their council seat, the decision as to who should succeed to that seat is taken by the party according to its own rules. When an Independent councillor resigns their seat, the decision as to who should succeed is taken according to the local authority’s Standing Orders and varies from council to council.

The TDs vacating the council seats post 2003 had several months in which to do so, giving most of them adequate leeway to ensure that their chosen candidate would succeed them.

In very many cases that chosen successor was a family member. Thus the observable implication of a dramatic increase in the rate of prior council service among family followers since 2003 strongly supports the Theory of Dynasty Formation.

5.8.3: Alternative Hypotheses:

Three of the alternative hypotheses identified in Chapter 2 as possible explanations for variations in the rate of family following are testable at a sub-national level. Testing them against the results of the last 11 Irish general elections yields the following results:

182

Alternative Hypothesis 3: That family following, even within mature developed, democratic states is linked to underdevelopment and thus will be more prevalent in less urbanised areas. Differences in the rate of family following between countries may thus be a function of the proportion of MPs in each parliament who represent rural areas .

If Putnam (1976) were correct and the singular prevalence of family seats in Ireland is a function of the semi-traditional nature of Irish society and if Mair (1981) is correct and

Ireland is divided between a traditional west and a modern east , then one would expect that more urbanised, affluent and developed parts of the country, such as liberal Dún

Laoghaire which bucked the national trend in voting for the divorce referendum in 1986 and against the pro-life amendment in 1983, would have few family seats, while more remote, poorer, rural areas like conservative Cavan-Monaghan which recorded the highest vote in the country in favour of the pro-life amendment in 1983 and the fourth highest against divorce in 1986 (DOEHLG 2009), would have more, at least in the early years of this case study. If so one would be quite wrong. Not once in any of the ten elections between 1981 and 2011 did the electors of Cavan-Monaghan return a TD who was related to a previous

Oireachtas member, though they did finally elect one family follower in 2016, grandniece of former TD Paddy Smith.

Sophisticated Dún Laoghaire, by contrast elected family TDs in eight of the nine elections between 1981 and 2007, though they have returned none in the last two elections. In 2002 three of the five Dáil seats in Dún Laoghaire were held by family TDs, the highest rate of family following in the entire country in that election. It should be noted that under the

183 methodology adopted in this study David Andrews TD was not counted as holding a family seat. This is despite the fact that his father Todd was one of the founders of Fianna Fáil and held a wide variety of government appointed posts, but Todd Andrews was never a member of the Oireachtas. If David Andrews were included in the list of family seats, Dún Laoghaire would comfortably emerge as the most dynastic constituency in the entire country. Of course selecting two constituencies like this is not statistically reliable, and other rural constituencies such as Kerry South have on occasion shown a marked propensity to return family TDs; but for Putnam's (1976) and Mair's (1981) points to be true, we should see clear regional differences in the proportion of family seats, with Dublin as the wealthiest, most populous, urbanised and most developed part of the country returning relatively fewer family seats than the rest of the country.

Graph 33: Family Seats by Region 1981 - 2016

35.00%

30.00%

25.00% Munster 20.00% Leinster (ex Dublin) Dublin 15.00% Connaught/Ulster

10.00%

5.00%

0.00% 1981 1982 1982 1987 1989 1992 1997 2002 2007 2011 2016 – Feb – Nov

184

As the graph above shows, this was emphatically not the case until the collapse of the

Fianna Fail vote in 2011. That collapse was particularly severe in Dublin where the party retained only a single seat. That TD, was a family follower, but seven other

Fianna Fáil family followers who won seats in Dublin in 2007 were not returned in 2011. In only one of the nine elections in this study prior to 2011 did Dublin return a lower proportion of family seats than the rest of the country and even then it was at a level that had been seen before elsewhere. In four of those nine elections Dublin returned the highest proportion of family seats, while in four others it returned the second highest.

Only since 2011 has the order that might have been expected from Mair's (1981) analysis, a clear gap between the West of Ireland comprising of Connaught/Ulster and Munster and the East, Dublin and Leinster, been achieved. As Fianna Fail recovered partially in 2016, the rate of family following in the East rose at a faster rate than in the West, and while it is too early to predict a reconvergence between the regions, the evidence does not support

Alternative Hypothesis 3 over the longer term. These results instead suggest strongly that family seats are a function of the electoral system and not of a traditional rural culture.

Alternative Hypothesis 4: That the rate of family following is simply a function of the superior quality of candidates from political families, in particular their higher levels of education and that therefore family followers will be more likely to be graduates.

185

Graph 34: Graduates 1981 - 2016 90.00%

80.00%

70.00%

60.00%

50.00% Family Followers Non Followers 40.00%

30.00%

20.00%

10.00%

0.00% 1981 1982 1987 1992 1997 2002 2007 2011 2016

As noted in Chapter 4 Ireland is the only country in the broader study that gives support to

Hess’ (1997) notion that family followers are better educated than non-followers and that it is their superior educational achievements that results in their greater electoral success.

The first part of the equation is certainly true. Throughout the period under study, family followers have been better educated than non-followers. As pointed out in Chapter 4, Irish family followers are not better educated than MPs in other parliaments. The anomaly in

Ireland is the poor level of education of TDs in general.

Most TDs enter the Dáil for the first time in their 40s some decades after finishing their formal education. The key factor in shaping the number of graduates available to be elected to the Dáil may then be considered to be the ease of access to 3rd level education in Ireland

186 in the twenty years prior to the period under study, effectively from 1961 to 1998. In the relevant decades before the beginning of this study, access to higher education in Ireland was limited, as was the type of high paying jobs, such as that of TD, which could enable a family to send their children to university. In such circumstances a gap between the proportion of graduates among the close relatives of TDs and that among other members of parliament is perhaps to be expected.

That gap narrowed somewhat in the early years of this study, again as would be expected with the combination of a rapidly improving economy and a dramatic increase in access to higher education. By 1997 the gap was less than 10 percentage points and the utility of the education gap in explaining the sustained high rate of family following in Ireland declined.

In recent elections however, far from narrowing, the gap has grown wider. This appears to be directly related to changes in the party system. There is a clear educational divide between the parties. 100% of the Labour party TDs in the current Dáil are university graduates, as are 61% of Fianna Fáil TDs and 56% of Fine Gael TDs. By contrast 45% of

Independent TDs and just 26% of Sinn Féin TDs are graduates. Therefore the rise of Sinn

Féin, and to a lesser extent Independents, in the last two elections and the decline of Fianna

Fáil in 2011 and Labour in 2016 appear to be the proximate causes of the declining number of graduates in the Dáil.

This has happened before. The proportion of all TDs who are graduates rose steadily from

1981 until 2002 when it suffered a sharp but temporary decline, coinciding with an equally sharp and temporary rise in the number of first time, localist Independents elected that year

187 as a result of Fine Gael’s electoral collapse. As the party system re-established itself in 2007, the proportion of graduates rose again. The 2011 and 2016 results by contrast show a steeper and more sustained decline in the number of graduates, coinciding with both a resurgence in the number of Independent TDs and Sinn Féin’s emergence as the third party in the Dáil.

Within Sinn Féin there is something of a generational divide. Many of the party’s younger

TDs are graduates while few of the older generation are. Over time therefore the proportion of graduates among Sinn Féin TDs, and therefore in the Dáil itself should begin to rise again. Whether this results in a decline in the education gap between family followers and non-followers will largely depend on how many of these new Sinn Féin graduates are family followers While Sinn Féin, as of yet, has only a single family follower in Dáil Éireann, it is to be expected that as the party’s tenure as a significant Dáil presence extends, and as it runs more candidates and thus experiences more intra-party competition, that more political dynasties will emerge. For instance it is widely expected that Cllr Toiréasa Ferris, a graduate of the University of , will run for the seat held by her father Martin at the next general election and should she be successful, increase the proportion of both graduates and family followers among Sinn Féin TDs.

Alternative Hypothesis 4 has to therefore be considered a possible explanation for the success of family followers in Ireland, though the fact that decline in the education gap prior to 1997 was not mirrored by a decline in the rate of family following weakens it somewhat.

188

Alternative Hypothesis 5: That the rate of family following is linked to traditionalism and will therefore be higher among socially conservative parties regardless of the electoral system.

Graph 35: Family Following by Ideology

35.00%

30.00%

25.00%

20.00% Socially Conservative Parties & Independents 15.00% Other Parties & Independents

10.00%

5.00%

0.00%

As can be seen from the graph above, the proportion of family followers in socially conservative and socially liberal parties was broadly similar for the first four elections in the period under study. Since 1989 however, there has been a marked divergence. Family following in Ireland is now very much associated with the more socially conservative parties,

Fianna Fáil and Fine Gael. The idea therefore that there is a link between family following and social conservatism does receive support from the Irish case, though not, as has been seen from the cases of Canada, Australia, New Zealand, Belgium and the USA, while the aggregate multivariate analysis in Chapter 4 actually showed a small negative correlation between social conservatism and family following. In the Irish case it should be pointed out that most socially liberal parties run only a single candidate in each constituency. It is the main socially conservative parties, Fianna Fáil and Fine Gael, who experience the full impact of intra-party competition in Ireland and whose candidates thus have the strongest

189 incentive to create and maintain a personal political machine. In this regard both Theory of

Dynasty Formation advanced in this thesis and the hypothesis that family following is linked to social conservatism as suggested by Doyle et al. (2015) predict the observed results.

Four additional points might be made in support of the idea that it is intra-party competition rather than social conservatism that is the key driver of family following in Ireland. Firstly, as observed above, there is no negative correlation between family following and urbanism.

If socially liberal Dún Laoighaire is more likely to elect family followers than socially conservative Cavan-Monaghan, then the link between family following and social attitudes is not clear. Secondly, the rate of family following among left wing TDs rose in 1992 when the Labour party – anticipating a good election performance – ran additional candidates thereby increasing their rate of intra-party competition but not changing their liberal stance on social issues. The Theory of Dynasty Formation would predict such an outcome. The idea that social conservatism drives family following would not. Thirdly, as explained above, because of the relative youth of the party in Dáil terms, Sinn Féin has yet to develop a significant number of political dynasties. As Sinn Féin is now the largest socially liberal party in the Dáil, this has a dampening effect on the overall rate of family following among such parties. Finally, both Fianna Fáil and Fine Gael became more socially liberal over the period under study. It was Fianna Fáil led governments who legalised contraception, decriminalised homosexuality and abolished the concept of illegitimacy. Fine Gael led governments liberalised the law on contraception further, introduced civil divorce, same sex marriage and held the referendum that repealed the constitutional ban on abortion. In neither party has this growing social liberalism been mirrored by a decline in family

190 following. It should also be noted that the Progressive Democrats, while economically conservative, were socially liberal on all of these same issues and at their intiation even proposed removing all references to God from the Irish Constitution, a socially radical proposal in 1980’s Ireland from a party with a high number of family followers among its

TDs.

5.9 Additional Findings:

The two additional factors looked at in the cross-national study, gender and average age at first election, have also been looked at in this longitudinal study. Neither can explain variations in the rate of family following between countries, but both are relevant when discussing the wider effects of political dynasties.

Graph 36: Average age at 1st Election

55 53 51 49 47 Family Followers 45 Non Followers 43 41 39 37 35 1981 1982 1982 1987 1989 1992 1997 2002 2007 2011 2016 Feb Nov

Like their counterparts in Canada, Netherlands and Belgium and unlike those in New

Zealand, Australia and the United States, Irish family followers are notably younger than

191 non-followers with an average age gap over the eleven elections of more than four years and one month At no point in those last eleven elections have newly elected followers been older on average than newly elected non-followers. The gap narrowed in 1997 to one year and eight months before widening again to its largest ever margin of eleven years and five months in 2007. While the average age of all TDs has fallen sharply since 2011, the average age of family followers has also fallen and they remain comfortably the younger cohort.

Graph 37: Women TDs as a % of Family Followers and non-Followers 30.00%

25.00%

20.00%

Family Followers 15.00% Non Followers

10.00%

5.00%

0.00% 1981 1982 1982 1987 1989 1992 1997 2002 2007 2011 2016 Feb Nov

The story of female family followers in Ireland is one of two eras. Before 2011 women made up a very small proportion of the Irish Dáil but a more respectable proportion of family followers. Since 2011 the opposite is the case. The level of female participation among

192 non-family followers more than doubled over the last two elections to over 25%, while the proportion of female family followers collapsed to zero in 2011 before rebounding to just

10% in last year’s general election.

As we have seen, in all countries in the cross-national study except the United States, family following is a decidedly male affair, as it now appears to be in Ireland. Even the relatively better performance of female family followers before 2011 should not be overstated. The average rate of female family following between the general elections of 1981 and 2007 was just 17.1%. Given that TDs will generally have an equal number of male and female relatives to whom they could bequeath their personal political machines, any idea that family following empowered female participation, even in these years, does not stand up to scrutiny.

Why the number of female family followers should have collapsed so completely in 2011 is a little surprising. The electoral collapse of Fianna Fáil was clearly the major contributing factor. All four of their female family followers either retired or lost their seats. Labour’s sole female family follower from 2007, Mary Upton also retired as did Olwyn Enright of Fine

Gael. That left just a single sitting female family follower to contest the election for the two parties, Fine Gael and Labour, who gained most seats. That candidate, , lost her seat to a running mate from her own party. By contrast the number of male family followers in the Fine Gael parliamentary party in 2011 jumped from 12 to 15 and the number of male family followers in Labour’s Dáil party increased from 2 to 4.

193

5.10 Conclusions:

In conclusion those elements of the hypotheses and additional observable implications that are testable at a sub-national level are all supported by the evidence of the last 11 Irish general elections.

Of the alternative hypotheses that are testable at a sub-national level, one, that of a rural bias towards family following is not supported by the evidence over the longer term. There is support for the proposition that family followers are more likely to represent socially conservative parties, but given the dominance of such parties in Irish politics over the period and that these are the only parties to experience fully fledged intra-party competition, that finding is not unexpected and does not undermine the Theory of Dynasty Formation. The lack of evidence in support of the idea of social conservatism as the prime cause of family following from the other countries in the cross-national study is of particular note.

Of greater interest is the much higher rate of educational achievement among family followers. This may not be sufficient to explain the consistently high rate of family following between 1981 and 1997, a period in which the educational gap actually narrowed but it is an explanation that may have some overall merit in the Irish case, though, like social conservatism, it does not appear to be a factor in the other countries in this study. Certainly the fact that Irish TDs are so poorly educated compared to their peers in other parliaments, and that the problem has gotten worse over the last two elections is, at the very least, something that is worthy of broader study.

194

It remains the case that Ireland’s rate of family following between 1981 and 2016 was, as the Theory of Dynasty Formation predicts, at a very high level when compared with the known results from other countries. This was a period of quite dramatic social, economic and political change. In light of such volatility, the consistency of family representation is striking. It is however part of an even longer pattern. Brian Farrell (1970, 1974) identified

40 TDs elected in 1969 and 45 in 1973 as being in a “close family relationship with former ministers, deputies and senators”. That the rate of family following in Ireland should stay so robust for so long is strong evidence for a system based explanation such as that provided by the Theory of Dynasty Formation outlined in this study rather than it being a purely cultural phenomenon.

From the information gathered on members of Dáil Éireann elected since 1981 a picture emerges of the typical newly elected member of an Irish political dynasty. He - and it will usually be a he – is the son of a former Fianna Fáil or Fine Gael TD. He lives in the same town, village or urban ward as his father and has served on the local city or county council since taking over his father’s council seat some years before. He concentrates on local concerns and serving the individual needs of his constituents. His career prospects are excellent. He stands a very good chance of early promotion to a junior ministry or even to cabinet. He is a university graduate and may well have worked in an established profession most likely as a primary or secondary school teacher before taking up politics fulltime, but his real job was always to nurture and maintain the personal political machine which delivered electoral success to his father. That machine now serves his electoral interests and in the future, he hopes, it will serve the interests of his son.

195

Chapter 6: Summary and Conclusion

6.1 Summary:

Family following in democracies has traditionally been seen as a function of underdevelopment (Putnam 1976; Camp 1982; Riedinger 1996; Schatz 2004; Besley 2005;

Acemoglu and Robinson 2006; Rossi 2009; Chhibber 2011), yet family following in electoral politics persists even in developed states (Laband and Lentz 1985; Van Liefferinge and

Steyvers 2007; Dal Bó et al. 2009; Kato S 2009; Smith 2012; Lumb 2012; Godwin 2013; Van

Coppenolle 2014; Doyle et al. 2015). Quite significant variations exist in the observable rate of family following among these mature developed, democratic states from lows of 2% in the Netherlands to a high of almost 20% in Ireland. It is that variation that this thesis seeks to explain.

The key contribution to this field of study was made by Daniel Markham Smith (2012).

Smith ranked eight democracies according to the degree to which their electoral systems were “candidate centered”. He studied the results of multiple elections in those democracies over several decades and, in seven of his eight cases, established a clear correlation between the degree to which electoral systems are candidate centered and the rate of family following. Smith’s theory however held that local control of the nominating process was an equal factor to the candidate centeredness of the electoral system and this hypothesis was only supported by the evidence from a single country, Iceland, the only country whose findings did not support his electoral system hypothesis.

196

This study takes a different approach. It proposes a mechanism of political inheritance based on the inheritance of personal political machines. Machines are mentioned elsewhere in the literature (Kato S 2009; Smith 2012). What this study does differently is that it defines the elements of an electoral system that incentivise the creation and maintenance of such personal machines and posits a Theory of Dynasty Formation that argues that there is a direct causal relationship between the incentive to create and maintain a personal political machine fostered by an electoral system and the rate of family following.

This thesis then proposes a new ranking of electoral systems according to the degree to which they possess the elements that incentivise the creation and maintenance of personal political machines. This ranking differs from previous rankings of the openness of electoral systems (Carey and Shugart 1995; Shugart 2001; Farrell and McAllister 2006; Colomer 2011) in that its measures are candidate, not voter centered. For example, an electoral system such as the Alternative Vote, which rates very highly for openness (Colomer 2011), is ranked quite low in terms of the degree to which it incentives the creation and maintenance of personal political machines.

Unlike previous cross-national studies (Smith 2012; Doyle et al. 2015), a systematic approach to case selection is taken which focuses solely on mature, developed, democratic state, with low levels of perceived corruption, excluding micro-states. A single recent election is studied for seven countries, each one representing a type of electoral system

197 from the ranking. In addition a longitudinal study of Ireland is undertaken to test the hypotheses over time.

This study records a wider range of attributes than previous cross-national efforts namely gender, age, education, prior experience in local and regional government; whether the constituency the MP represents is urban, rural or mixed and whether the party they represent is socially conservative or liberal and whether it is traditionally successful. It also proposes a clearer distinction between different forms of inheritance: direct same constituency, delayed same constituency and extra-constituency. These attributes are used to test the four hypotheses, six additional observable implications and five alternative hypotheses as summarised below.

6.2 Findings Recap:

The Theory of Dynasty Formation outlined in Chapter 2 states that the key causal variable in determining variations in the rate of family following between mature, developed, democratic states is the degree to which the electoral system incentivises the creation and maintenance by MPs of personal political machines. Three elements of an electoral system were posited to increase the incentive to create and maintain such personal political machines and therefore to lead directly to higher rates of family following. These are: personalised voting, intra-party competition and preference vote seeking from other parties in the presence of intra-party competition.

This thesis led to four testable hypotheses. They were that:

198

1. Political Dynasties will be more prevalent in electoral systems that allow for

personalised voting

2. Electoral systems that allow for intra-party competition will have much higher rates

of family following than those that do not;

3. Electoral systems that allow for preference vote seeking from other parties but do

not allow for intra-party competition will not demonstrate notably higher rates of

family following;

4. By contrast, electoral systems that allow for intra-party competition and preference

vote seeking from other parties will will show the highest rate of all.

In addition, if the theory held, six further observable implications should be clear, namely:

I. That because what is being inherited is a personal political machine, then instances

of family following under systems that incentivise the creation and maintenance of

such machines will tend to be in the same constituency

II. Similarly instances of family following under systems that incentivise the creation

and maintenance of personal political machines will tend to happen immediately

following the death, defeat or retirement of the preceding family member before

the machine has time to decay.

III. That in electoral systems that allow for intra-party competition, family following will

be more prevalent among more traditionally successful parties including among

former members of such parties who choose to run as Independents.

IV. That in electoral systems that allow for intra-party competition family followers will

be disproportionately represented among high office holders such as cabinet

ministers.

199

V. In countries with mixed systems such as MMP, the rate of family following among

constituency MPs will be higher than among list MPs.

VI. The proportion of MPs with prior experience in elected office at local or regional will

increase as the the degree of intra-party competition increases.

As seen in Chapter 4 all four hypotheses and all six additional observable implications are strongly supported by the results of recent elections in the seven mature, developed democracies in this study. In addition, those elements of both the hypotheses and the additional observable implications that are testable at a sub-national level are supported by the results of longitudinal study on Ireland in Chapter 5.

The evidence in support of the Theory of Dynasty Formation is strong and the theory itself, on the basis of the evidence, is deemed to be supported.

From a review of the literature on family following, six alternative hypotheses were advanced that might offer an alternative explanation for variations in the rate of family following between mature, developed, democratic states, and in doing so challenge the

Theory of Dynasty Formation.

These were:

1. That family following, even in mature, developed, democratic states is linked

to rent seeking behaviour.

2. That family following, even in mature, developed, democratic states is linked

to underdevelopment.

200

3. That family following, even within mature, developed, democratic states is

linked to underdevelopment and thus will be more prevalent in less

urbanised areas.

4. That the rate of family following is simply a function of the superior quality of

candidates from political families, in particular their higher levels of

education.

5. That the rate of family following is linked to traditionalism and will therefore

be higher among socially conservative parties regardless of the electoral

system.

6. That the rate of family following is a function of the degree of electoral

system openness and not of intra-party competition.

None of these six alternative hypotheses could successfully account for the variations in the rate of family following found in the comparative cross-national element of this study and all six are held to fail on that basis alone. Alternative hypotheses 4 and 5 do correlate with the findings in the longitudinal Irish study, but in both instances the Theory of Dynasty

Formation offers a stronger explanation of the observed results. Nevertheless the negative correlation between the high rate of family following in Ireland and the low education level of non-family followers in the Irish Dáil is deemed worthy of further study.

6.3 Further Research:

6.3.1 More elections:

It would be advisable to contrast the findings of this study with the results of more recent and upcoming elections in each of the sample countries. The rate of family following as

201 found in this study is consistent with that found in studies of Canada (Godwin 2013), the

United States (Dal Bo et al. 2009) and Australia (Lumb 2012; Lumb and Gobbett 2015) and with Smith’s comparative cross-national study (Smith 2012). Nevertheless, one of the principals of the Theory of Dynasty Formation outlined in chapter 2 is that the rate of family following should be reasonably consistent over time. Clearly an unusual election result, as occured in Ireland in 2011, can have a short term impact, but that should balance out as more elections are included, as indeed appears to be happening already in the Irish case. A second reason is to increase the sample size to improve the robustness of the conclusions.

This is particularly relevant in the case of MMP in New Zealand where despite the apparent clarity of the result, the sample size of both list and constituency MPs are simply too small for the findings to be entirely trusted.

6.3.2 More countries:

Just one example for each electoral system was looked at in this study. Increasing that number should improve overall reliability. While a second country may just illustrate that one of them is an anomalous choice, it will increase the sample size for each system and should have the effect of increasing overall robustness. Three or more countries would be better, but some electoral systems such as Alternative Vote and PR-STV are not used widely enough for that. A list of countries that would fit the parameters could include Austria for

Closed-List PR, the United Kingdom for SMP, France for SMM – which should be broadly comparable with AV - Germany for MMP, or any of a wide range of Western European states for Open-List PR. Malta, the only other country that uses PR-STV falls outside the parameters of this study on population size and level of development, but a comparative longitudinal study of Ireland and Malta may also be of value.

202

6.3.3 Gender:

The question of a gender imbalance in favour of men among family followers has not been previously identified in the literature, but is clearly present in the data. Smith (2012) and

Folke et al. (2016) have both found that historically women were more highly represented among family followers than non-followers, a phenomenon also found in the longitudinal study of Ireland in Chapter 5. In Ireland this relative bias towards women disappeared in

2011 and in the cross-national study women form a smaller proportion of family followers than non-followers in every country except the United States.

Rossi (2009) did find that family followers in Argentina were disproportionately likely to use the same surname as their predecessor. This may perhaps be taken as a proxy for gender.

The explanation advanced by Rossi was one of the surname serving as a brand. Support for this may come from the findings on the United States. As noted, the US is the only country in this study where women are more prevalent among family followers than non-followers and in the US, all bar one of the female family followers uses the surname of her male predecessor.

Given that family followers are elected at a younger age and are promoted to senior positions more easily, the lack of women among them raises serious questions for gender equality. Simply increasing the number of women in parliament may not be enough to counter a power imbalance arising from male domination of political dynasties. In Ireland the 2016 election was the first one held using candidate gender quotas. It resulted in a record high number of women TDs being elected. This was reflected in the make up of the

203 cabinet formed that year which for the first time ever was over 26% female. However family followers, all of whom were male, made up almost 47% of that same cabinet, a hereditary hold on power of quite stunning proportions.

6.3.4 Why poor level of education in Ireland?

A final area for further research may be a parochial one, but the reasons for the very low, and declining, levels of educational attainment among Irish TDs compared to their peers needs to be comprehensively established. Irish TDs already spend a disproportionate amount of time on constituency work. This has a real effect on the time they can allocate to parliamentary oversight and legislative work (JCC 2010). Added to this, the fact that so few have the qualifications or professional knowledge to either hold senior civil and public servants to account or to contribute adequately to the formation of legislation is a real concern. The financial crash that bankrupted Ireland in 2008 was caused, at least in part, by a lack of effective oversight at a national level. Recent revelations about the handling of data by , a matter raised with the Irish Data Commissioner in 2013 (Edwards 2017), have again shone an unflattering light on the ability of the Irish state to oversee complex operations.

The problems caused by inadequate oversight were implicitly recognised by former

Taoiseach Enda Kenny when he forbade the ministers in his incoming cabinet from engaging in constituency work, an instruction which, given the highly localised nature of Irish politics, was more honoured in the breach than the observance (McEnroe 2011).

204

6.4 Political dynasties in democracies matter

As we have seen, a significant number of parliamentarians across a range of mature, developed, democratic states are drawn from political dynasties. These family followers are far more likely to be men. They also enjoy a much greater likelihood of promotion to high office. Ministers are more likely than MPs to be members of dynasties. Heads of government even more so.

While judging the effectiveness of elected politicians is not easy, the evidence from the business world is clear. Dynasties hinder rather than help good governance. It should perhaps be no surprise that although, as seen in Chapter 4, the rate of perceived corruption in a country is not a causal factor in the rate of family following, it remains a fact that the four countries in the cross-national study with the lowest rates of perceived corruption –

New Zealand, Canada, Australia and the Netherlands, are also the four with the lowest rate of family following.

Even if dynasties themselves cause no direct harm, in an age when much of politics is concerned with the apparent disconnect between voters and the political class, between rulers and ruled, the fact that there are so many hereditary politicians at the very apex of political power in the most developed countries on earth is surely worthy of notice and of concern.

6.5 Measures to reduce the prevalence of political dynasties:

Reducing the prevalance of dynasties is not easy. The Theory of Dynasty Formation is quite clear, the rate of family following is hard wired into the electoral system. Among developed

205 countries only Japan, by changing its electoral system to one with less intra-party competition, has succeeded in achieving a significant reduction in the rate of family following in recent years. Changing an electoral system is expensive and difficult and, as the

Japanese example shows, the benefits can take several election cycles to accrue (Smith

2012).

A more extreme option might be to ban dynasties altogether. Such a law was passed in the

Phillipines decades ago but never implemented (Riedinger 1995). In Japan the opposition

Democratic Party proposed a ban on dynasties in 2009, a proposal somewhat marred by the fact that the party leader was himself a member of a political dynasty. Apart from its impracticality, banning persons from standing for election because of who they are related to is inherently undemocratic and unlikely to receive public support.

Measures to neutralise the advantage conveyed by inherited political machines are problematic. Such measures could include enhanced public funding or guaranteed access to public airwaves for non-dynastic candidates but these are unlikely to meet with widespread approval. Even in the most dynastic country in this study, Ireland, most candidates for office are not family followers. A system that tries to enhance the electoral chances of the vast majority of candidates is likely to be expensive, complicated and ineffective.

If the theory is correct, the key to reducing the prevalence of dynasties is to reduce the rate of intra-party competition and thereby the incentive for politicians to create and maintain a personal political machine. Approaches to achieving these goals will vary but for the three

206 countries in this study with the highest overall rates of family following the following measures might have a positive impact.

United States: Intra-party competition in the US occurs at the candidate selection stage and is exacerbated by the problem of gerrymandering. Many Congressional districts have such large in-built majorities for one party or the other that the general election is seen as a foregone conclusion. Some of these constituencies are of a bizarre size and shape and clearly constitute an attempt by politicians to undermine the democratic process (Mann

2016). In these circumstances, the competition for a party nomination that all but guarantees a seat in the House of Representatives is likely to be particularly fierce. In simple terms a nomination to a “safe seat” is more valuable than a nomination to a competitive seat and far more valuable than a nomination to a hopeless seat. It should therefore follow that safe nominations attract a greater level of intra-party competition in the same way that a higher than normal price level will attract more entrants to a market.

This is precisely what Hirano and Snyder (2012) found. According to their research some

60% of US Congressional districts are uncompetitive at the general election stage. These safe districts consistently attract significantly greater levels of competition at the primary stage than competitive districts do and incumbent members of Congress in such seats are more likely to lose to a party colleague in the primary than to an opponent in the general election (Hirano and Snyder 2012). In terms of the Theory of Dynasty Formation therefore there is a clear correlation between gerrymandering and safe seats, between safe seats and higher levels of intra-party competition and between higher levels of intra-party competition and political dynasties.

207

The most effective way of reducing the prevalence of such dynasties therefore may be to appoint, as in most modern democracies, an independent constituency commission to draw fair districts. Not only should this result in less fierce competition at the primary stage and thus in fewer political dynasties, over time it should also result in less partisan politics as candidates for election appeal to the centre ground of the electorate rather than the extremes within their own parties, something that can only benefit the political health of the United States.

Belgium: Belgium’s complicated political ecosystem is a contributing factor to the personalism of its politics, but ending the dual and triple mandates held by Belgian politicians is unlikely to reduce the rate of family following. Ireland banned members of the

Oireachtas from holding seats on local councils from 2003, but far from reducing the importance of dynasties it may actually have entrenched them. Many TDs and Senators, anxious to secure their local electoral base, now hand their vacated council seats to relatives. For example of the six councillors in Cork who were first elected to Dáil Éireann in

2016, three ensured that their council seats were taken over by siblings (O’Riordan 2016).

Reducing the number of family followers in Belgium may instead require electoral system tweaking rather than outright system change. Most forms of Open-List PR impose restrictions on the number of preference votes that can be expressed within a party list.

Imposing such a limit does not eliminate intra-party competition, or unduly affect voter choice but it does reduce the likelihood that a candidate will benefit from adopting a personal vote seeking strategy and therefore reduces the incentive to create and maintain a

208 personal political machine. Changing to something similar to the three preference votes allowed within a list in Sweden might be the most effective way of achieving this.

There is an argument that, as the Dutch experience shows, a further reduction in the incentive could be achieved by creating larger parliamentary constituencies. The counter argument in Belgium is that this was already tried in the 2003 reforms which reduced the number of constituencies from 20 to 11, though that reform was done alongside the introduction of a 5% threshold for winning parliamentary seats (Hooge et al. 2003) thereby weakening the smaller parties, who engage in less intra-party competition and benefitting the stronger parties, who engage in more. The effect of these two reforms was to see an increase in the rate of family following in Belgium (Van Liefferinge and Steyvers 2007).

The key therefore for Belgium in maintaining effective voter choice while reducing the electoral system effects that give rise to political dynasties, may lie in reducing the absolute degree of choice that voters now hold, by imposing some restrictions on their ability to rank candidates.

Ireland: In Ireland’s case larger constituencies, by promoting greater levels of intra-party competition, are more likely to exacerbate the problem than solve it. Reducing the size of constituencies is not practical either without abandoning any attempt at achieving a proportional election result, something with which PR-STV already struggles.

One possibility, discussed at the Joint Oireachtas Committee (JCC 2010i) would be to introduce above the line voting, giving voters the option to treat PR-STV like a form of

209

Closed-List PR. Such above the line voting is used in conjunction with PR-STV in Senate elections in Australia. Some 98% of Australian Senate voters chose to vote above the line, effectively eliminating intra-party competition in these elections, and reducing the proportion of dynasties in the Australian Senate to negligible levels. It should be noted that

Australian Senate elections take place in vast geographical constituencies where personalised vote seeking would be extremely difficult.

The Committee also considered changing the ordering of candidates on the ballot paper in line with the approach taken in Malta. There candidates are ranked alphabetically within party groups rather than in strict alphabetical order. This might well lead to a greater emphasis on political parties to the detriment of Independents, but it is difficult to see how such a ballot change would reduce intra-party competition. Indeed it should be noted that

Malta has a tradition of electing political dynasties on a level similar to that seen in Ireland; families such as the Mifsud Bonnici, Bonnici, Farrugia, Dalli, De Marco and Fenech Adami have dominated Maltese politics for generations. As discussed in Chapter 3, the country does not qualify as a mature, developed, democratic state, but further research into the effect of adopting the version of PR-STV used in Malta would be advisable. Whether Irish voters, or more pertinently, Irish politicians, could be nudged into changing their habits either by the appearance of an above the line option on the ballot paper or by the placing of candidates in party goups is debatable to say the least.

A more direct approach might be to try and break the link between local and national politics by imposing a bar on sitting or recently retired city and county councillors standing for election to the Dáil. The fear of losing one’s seat to a running mate who is a councillor is

210 one of the main reasons both for the tendency of TDs individually to insert family members into local authority seats in their own electoral bailiwicks and for TDs collectively to keep local government in Ireland weak and underfunded. Under the current system, for TDs to empower local councillors would be for them to empower their own future competitors and in effect, write their own political death warrants. If however councillors could not run for national office, say for a period of five years after vacating their council seats, this would facilitate the transfer of powers and functions to local government. This in turn should reduce the demand from the public for the constituency service provided by TDs and therefore, at least to some extent, the advantage conveyed by inheriting a political machine.

Even if the impact on the rate of family following of such a move were small, it may have the effect of ameliorating one of the worst effects of PR-STV – the excessive local focus it requires of national politicians. Such a temporary bar would not be as invidious to individual and voter choice as the sort of lifetime ban on dynasties suggested in Japan or the

Philippines. Similar temporary prohibitions exist worldwide against civil and public servants standing for office or retired politicians working as lobbyists.

Whether such a temporary bar would reduce the rate of intra-party competition is difficult to predict, but if along with the bar, measures were taken to make local government a more attractive career choice, for example by enhancing pay and powers of councillors and providing them with pensions, it is at least probable that more councillors would chose to remain at that level rather than abandoning their seats for a potential Dáil run five years hence. While the parties themselves would still seek to run multiple candidates, they would not be under the same degree of internal pressure as they currently experience from councillors to run a higher number of candidates. While such pressure is often resisted by

211 the parties, the resistance is not always successful and there are many examples, such as

Fine Gael in Limerick West in 1987, of a party running more candidates than was advisable to accommodate local councillors. A key reason why parties do sometimes give way to such demands is that not doing so could result in those same councillors, with their established local bases running as Independents, splitting the party vote and advising their supporters not to give preferences to the party’s candidates. This also happened in the same constituency of Limerick West, this time to Fianna Fáil in 2011 and has occured on other occassions throughout the country.

Some or all of these measures may help to reduce the rate of family following in Ireland, though it is probable that only an abandonment of PR-STV would bring it down to levels considered normal in other advanced states. Of the alternative electoral systems available most discussion in Ireland has focused on Mixed Member Plurality (JCC 2010i). The evidence from Japan’s experience and from this study would indicate that changing to MMP is likely to reduce the rate of family following over time. However when TDs discussed the possibility of adopting this system they expressed some serious reservations principally about the creation of two classes of TD, constituency and list, and the concern that the then dominant Fianna Fáil party would win the vast bulk of constituency seats, but few list seats, while for other parties the reverse would hold true (JCC 2010i).

A possibly more palatable alternative for Ireland might be to adopt the Alternative Vote (AV) system used in Australia. This should certainly have the effect of eliminating intra-party competition and thereby reducing the rate of family following. As the results of this study show, MPs elected under AV are almost 8 times less likely to be family followers than those

212 elected under PR-STV. AV is also a system that is familiar to Irish voters as it is used in all by- elections and Presidential elections in Ireland and for many internal elections for sporting and civil society groups such as the Gaelic Athelic Association and the Irish Farmers’

Association. From the voter’s perspective it is in effect the same system as PR-STV. Votes are cast in exactly the same way and counted in a manner that is familiar. It is of course not a proportional system, and this is likely to be the main argument against it, but unlike Single

Member Plurality, which Fianna Fáil led governments tried unsuccessfully to introduce twice in the past, elections under AV do consistently return small numbers of MPs who are

Independents or represent smaller parties and given its amenability to strategic voting, it is unlikely that a Dáil elected under such a system would be dominated by a single party for any prolonged period.

It would also have the effect of dramatically increasing the number of candidates on offer at election time to the public. In 2016 the four main Irish parties, Fianna Fáil, Fine Gael, Sinn

Féin and Labour ran a combined total of 255 candidates for the 158 seats in Dáil Éireann – an average of 1.6 candidates per seat, by far the lowest ratio of candidates to seats of any country in this study. AV, MMP, List PR and even SMP all lead to far more candidates being run. In any country struggling to bring talent into its parliament, retaining a system such as

PR-STV that so impedes the nomination of fresh blood must surely be questionable.

6.6 Conclusion:

Political dynasties in mature, developed, democratic states do not arise from corruption or underdevelopment, but from the degree to which the electoral system fosters intra-party

213 competition. As such they can be considered a negative externality arising from an understandable desire to employ an electoral system that maximises voter choice.

Reducing the prevalence of such dynasties while preserving a responsive and proportional electoral system is a difficult challenge, but it is a challenge that can be met. Understanding the factors that lead to heightened levels of family following is key. Voter choice alone is not the problem; excessive intra-party competition is. A highly open electoral system that does not facilitate intra-party competition, such as the Alternative Vote, will not foster undue dynasticism. Neither will a highly proportional system that does not facilitate intra- party competition, such as Closed-List PR. Attempting to square the circle of proportionality and openness runs the risk of fanning the flames of intra-party competition and embedding a dynastic elite. Like much in life openness and proportionality in moderation are beneficial. In excess they may lead to damage. Electoral reformers should proceed with caution.

214

References

Acemoglu, D and Robinson J (2006) 'Persistence of Power, Elites and Institutions' American Economic Review, American Economic Association: Washington D.C. Vol. 98 No. 1 Pages 267-293 March 2006

Altman, D, Chasquetti D (2005) ‘Re-election and Political Career Paths in the Uruguayan Congress’ The Journal of Legislative Studies Vol. 11 No.2

Amacher, R, Boyes, W (1978) ‘Cycles in Senatorial Voting Behaviour: Implications for the Optimum Frequency of Elections’ Public Choice 1978 Vol. 33 No. 3

Andersson, S and Heywood, PM (2009) ‘The Politics of Perception: Use and Abuse of Transparency International's Approach to Measuring Corruption’ Political Studies Volume: 57 issue: 4, page(s): 746-767

Ashworth, S de Mesquita, E B (2006) ‘Delivering the Goods: Legislative Particularism in Different Electoral and Institutional Settings’ Southern Association Vol. 68 No. 1 Pages 168-179

Bennedsen, M Nielsen, K Pérez-Gonsalez, F andWolfenson, D (2007) 'Inside the Family Firm: The Role of Families in Succession Decisions and Performance'The Quarterly Journal of Economics Vol. 122 Issue 2 Pages 647 – 691

Besley, T (2005) 'Political Selection' The Journal of Economic Perspectives Vol. 19 No. 3 Pages 43-60

Biglaiser, G, Mezzetti, C (1997) ‘Politicians Decision Making with Re-election Concerns’ Journal of Public Economics Vol. 66 No. 3

Bollen, K. A. (1986) Political rights and political liberties in nations: An evaluation of human rights measures, 1950 to 1984. Human Rights Quarterly, 8(4), pp. 567–591.

Bowler, S and Farrell, D (1991) 'Voter Behaviour under STV-PR: Solving the Puzzle of the Irish Party System' Political Behaviour Vol. 13 No. 4 Pages 303-320

Boyce, P.J. and Herr, R.A., (2008) ‘Microstate diplomacy in the south pacific’. Australian Outlook, (April 2012), pp.37–41.

Bradbury, J Russell, M (2005) The Local Work of Scottish MPs and MSPs: Effects of Non- coterminous Boundaries and AMs. Report to the Commission on Boundary Differences and Voting SystemsUniversity of Swansea: The Constitution Unit

Brooten, L (2013) ‘The Problem with Human Rights Discourse and Freedom Indices: The Case of Burma/Myanmar Media’ International Journal of CommunicationIssue 7 Pages 681- 700

Bullock, C 1976 ‘Motivations for US Congressional Committee Preferences: Freshmen of the 92nd Congress’ Legislative Studies Quarterly Vol. 1 No. 2 May 1976

215

Cain, B, Frerejohn, J, Fiorina, M (1984) ‘The Constituency Service Basis of the Personal Vote for US Representatives and British Members of Parliament’ The American Political Science Review Vol.78 Pages 110 -125

Camp, R. A (1982) 'Family Relationships in Mexican Politics: A Preliminary View' The Journal of Politics, Vol. 44, No. 3 (Aug., 1982), Pages 848-862

Carey, JM and Shugart, MS (1995) ‘Incentives to Cultivate a Personal Vote: a Rank Ordering of Electoral Formulas’ Electoral Studies Vol 14 No. 4 pp417-439

Carson, R (2007) On Rent Seeking Cost under Democracy and Carlton Economic Papers available: http://www.carleton.ca/economics/wp-content/uploads/cep07-01.pdf [accessed 20/11/14]

Carty, RK (1980) 'Politicians and Electoral Laws: An Anthropology of Party Competition in Ireland' Political Studies, Vol. XXVIII, No. 4 December 1st 1980 Pages (550-566)

Chhibber, P (2011) “Dynastic Parties: Organization, finance, and impact.” Party Politics 19(2) 277–295 available: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.936.9953andrep=rep1andtype= pdf [accessed 20/07/2017]

Clubok, A, Wilensky, N and Berghorm, F (1969) 'Family Relationships, Congressional Recruitment, and Political Modernization'The Journal of Politics, Vol. 31, No. 4 (Nov., 1969), Pages 1035-1062

Collins, S (2007) Nealon's Guide to the 30th Dáil and 23rd Seanad Dublin: Gill and Macmillan

Collins, S (2011) The Irish Times Nealon's Guide to the 31st Dáil and 24th Seanad Dublin: Gill and Macmillan

Collins, S (2016) The Irish Times Nealon's Guide to the 32nd Dáil and 25th Seanad Dublin: Gill and Macmillan

Colomer, J (2011) Personal Representation: the neglected dimension of electoral systems ECPR Press available: https://www.researchgate.net/publication/256238492 [accessed 01/10/2017]

Cross, W, Young, L (2008) ‘Factors Influencing the Decision of the Young Politically Engaged to Join a Political Party’ Party Politics Vol. 14, no. 3 pp 345 - 369

Dal Bó, E, Dal Bó, P and Snyder J (2009) 'Political Dynasties' Review of Economic Studies (2009) 76, Pages 115–142

Daniele, G. and Geys, B. (2014). Born in the Purple: Political Dynasties and Politicians’ Human Capital. Available: http://www.gianmarcodaniele.com/uploads/3/1/8/0/31808817/dynastic_politicians_and_p olitical_quality20140630_2.pdf [accessed 04/12/2015]

Dods (2010) Dod’s Parliamentary Companion 2011 London: Dod's Parliamentary Communications

216

DOEHLG (2009) Referendum Results 1937-2009 Department of the Environment, Heritage and Local Government: Dublin

Doyle at al (2015) Doyle, D McNulty, G Suzuki, A O’Malley, E ‘Democratic Dynasties: Explaining their prevalence in modern democracies’ Conference Paper PSAI Annual Conference, 2009 available: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2556830 [accessed 21/01/2017]

Djikstra L and Poelman H (2014) A harmonised definition of cities and rural areas: the new degree of urbanisation Regional Working paper 2014 available http://ec.europa.eu/regional_policy/sources/docgener/work/2014_01_new_urban.pdf [accessed 05/08/2017]

Dumienski, Zbigniew (2014). Microstates as Modern Protected States: Towards a New Definition of Micro-Statehood" Occasional Paper. Centre for Small State Studies. Available: http://ams.hi.is/wp-content/uploads/2014/04/Microstates_OccasionalPaper.pdf [accessed 08/12/2015]

Economist (2015) ‘Dynasties’ Economist April 18th 2015 available: http://www.economist.com/news/leaders/21648639-enduring-power-families-business- and-politics-should-trouble-believers [accessed 21/04/2015]

Edwards, E (2017) ‘All you need to know in the Max Schrems-Facebook case’ Irish Times February 6th 2017 available: https://www.irishtimes.com/business/technology/all-you- need-to-know-in-the-max-schrems-facebook-case-1.2965482 [accessed 28/03/18] EED (2017) European Election and Referendum Database Norwegian Centre for Research data available: http://www.nsd.uib.no/european_election_database/about/background.html [accessed 01/06/17] Ehin, P Madisse, U Solvak, M Taagepera, R, Vassil, K and Vinkel, P ‘Independent Candidates in National and European Elections’ 2013 available: https://www.ut.ee/sites/default/files/www_ut/uuringafco.pdf [accessed 18/03/18]

Elections Canada (2015) Elections Canada available: http://www.elections.ca/home.aspx [accessed 15/08/2015]

Enli, GS and Skogerbo E (2013) ‘Personalised Campaigns in Party-Centred Politics: Twitter and Facebook as arenas for political communication’ Information, Communication and Society Vol 16. No.5 July 2013 available: https://www.researchgate.net/publication/256568777_Personalised_Campaigns_in_Party- Centered_Politics_Twitter_and_Facebook_as_Arenas_for_Political_Communication_2015 [accessed 20/04/2018]

Epstein, L (1967) Political Parties in Western Democracies.New York: Praeger Publishers

Fanning, R (2011) 'Obituary: Declan Costello' 12/6/11 available: http://www.independent.ie/obituaries/declan-costello-2672723.html [accessed 18/8/11]

Farrell, B (1970) 'The 1969 Generation' in The Economic and Social Review edited by Bertram Hutchinson and John Bristow Volume 2 1970 – 1971 Pages 319-320

217

Farrell, D McAllister I (2006) ‘Voter Satisfaction and Electoral Systems: Does Preferential Voting in Candidate-centred Systems Make a Difference’ European Journal of Political Research Vol 45 Pages 723 – 749

Feinsten, BD (2010) The Dynasty Advantage: Family Ties in Congressional Elections Researchgate available: https://www.researchgate.net/publication/228256474 [accessed 05/09/2017]

Fenno, R (1977) ‘US House Members in Their Constituencies: An Exploration’ American Political Science Association Vol. 71 No. 3

Fiva, J Smith, DM (2016) ‘Political Dynasties and the Incumbency Advantage in Party- Centered Environments’ CesInfo Working Paper no. 5757 February 2016 available: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2747894 [accessed 03/08/2017]

Folke, O., Persson, T., and Rickne, J. (2015). Dynastic political rents. Columbia University. Mimeo. Available: http://perseus.iies.su.se/~tpers/papers/FPR_Dynastic_150730.pdf [accessed 12/07/2017]

Folke, O Rickne, J and Smith, D.M. (2016). ‘Gender and dynastic political recruitment: theory and evidence’, Mimeo, Uppsala University available: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2985230 [accessed 17/04/2018]

Freedom House (2016) Freedom in the World 2016 Anxious Dictators, Wavering Democracies: Global Freedom under Pressure available: https://freedomhouse.org/report/freedom-world/freedom-world-2016 [accessed 03/02/2017]

Gagliarducci, S Manacorda, M (2015). Politics in the family: and the hiring decisions of Italian firms. Mimeo, Queen Mary University. Available: http://ftp.iza.org/dp9841.pdf [accessed 20/06/2017]

Gallagher, M Komito, L (2009) ‘The constituency role of TDs’, in J. Coakley and M. Gallagher (eds), Politics in the Fifth Edition, Routledge, 2009, pp. 230–62

Gallagher, M (1988) Introduction. In M. Gallagher, and M. Marsh (Eds.), Candidate Selection in Comparative Perspective: The Secret Garden of Politics (pp. 1-19). London: SAGE Publications.

Gallagher, M (2000) 'The (relatively) victorious incumbent under PR-STV: legislative turnover in Ireland and Malta' in Elections in Australia, Ireland and Malta under the Single Transferable Vote: Reflections on an inbedded institution edited by Bowler, S and Grofman, B Pages 81 – 113 Ann Arbour: University of Michigan

Gallagher, M (2008) 'The Discreet Charm of PR-STV' in The Politics of Political Systems eds Gallagher, M and Mitchell P Pages 511-529 Oxford University Press 2008

Gelman, A King, G (1990) ‘Estimating Incumbency Advantage without Bias’ American Journal of Political Science Vol 34 No. 4 1990 available: www.jstor.org/stable/2111475. [accessed 02/03/2018]

218

Giannone, D. (2010). Political and ideological aspects in the measurement of democracy: The Freedom House case. Democratization, 17(1), 68–97.

Goodin, E (1982) ‘Rational Politicians and Rational Bureacrats in Washington and Whitehall’ Vol. 60 No. 1

Godwin, M (2013) Parliamentary Privilege? Kinship in Canada’s parliament Conference Paper PSA PGN Oxford December 2013

Government of Norway (2009) The main features of the Norwegian Electoral System – Summary available: https://www.regjeringen.no/en/topics/elections-and-democracy/den- norske-valgordningen/the-norwegian-electoral-system/Norwegian-Electoral-System- Summary/id570513/ [accessed 30/07/2017]

Heithusen, V Young, G and Wood, D (2005) 'Electoral Context and MP Constituency Focus in Australia, Canada, Ireland, New Zealand, and the United Kingdom' American Journal of Political Science, vol. 49 (1) (Jan., 2005), pp. 32-45

Hess, S (1997) America’s Political DynastiesNew Brunswick: Transaction Publishers

Hirano, S and Snyder, JM (2012) ‘Primary Elections and the Quality of Elected Official’ Harvard University and NBER available: https://www.vanderbilt.edu/csdi/events/Hirano.Talk.pdf [accessed 22/03/2018]

Holland, T (2004) Rubicon: The Triumph and Tragedy of the Roman Republic, London: Abacus

Hooge, M Noppe, J and Maddens, B (2003) ‘The effect of electoral reform on the Belgian election results of 18 May 2003’ Representation Vol 38 4 pp 270-276

IPU (2016) Inter-Parliamentary Union Parline Database available: http://www.ipu.org/parline-e/parlinesearch.asp [accessed 17/06/2016]

JCC (2010) Joint Committee on the Constitution Third Report Article 16: Results of Survey of Members of Both Houses of the Oireachtas: The Electoral System, Representative Role of TDs and Proposals for Change Interim Report Dublin: Houses of the Oireachtas

JCC (2010i) Joint Committee on the Constitution Fourth Report Article 16 of the Constitution: Review of the Electoral System for the Election of Members to Dáil Éireann Final Report Dublin: Houses of the Oireachtas

Kalt, J Zupan, M (1984) ‘Capture and Ideology in the Economic Theory of Politics’ The American Economic Review Vol. 74 No. 3

Kato, H (2009) 'Rebuilding Japanese Politics by Establishing Self-governance in Political Parties: A Challenge for the New Administration' Insights into Japanese Politics and Society Pages 7 – 22 Toyko: The Tokyo Foundation 2009

Kato, S (2009) 'Hereditary Lawmakers in an Era of Politically Led Policymaking' Insights into Japanese Politics and Society Pages 87 – 100 Toyko: The Tokyo Foundation 2009

219

Kavanagh, A (2016) Does size matter? Largest and smallest General Election 2016 constituencies (in terms of area)available: https://adriankavanaghelections.org/2016/01/20/does-size-matter-largest-and-smallest- general-election-2016-constituencies-in-terms-of-area/ [accessed 01/12/2016]

Kavanagh, A (2018) Looking at the Spring 2018 polls: Constituency-level analysis of the 26th March Sunday Business Post-Red C opinion poll and other polls available: https://adriankavanaghelections.org/2018/03/26/looking-at-the-spring-2018-polls- constituency-level-analysis-of-the-26th-march-sunday-business-post-red-c-opinion-poll-and- other-polls/ [accessed 28/03/2018]

Klein, E and Umit, R (2016) ‘The electoral connection of ministerial selection in the UK’ Journal of Legislative Studies Vol 22 No. 2 available: https://www.tandfonline.com/doi/abs/10.1080/13572334.2015.1134904 [accessed 24/05/18]

Knesset (2014) Family ties between Knesset members available: https://knesset.gov.il/mk/eng/family_eng.asp [accessed 07/01/2014]

Kurtz, D (1995) 'Inheriting a political career: The justices of the United States and Louisiana Supreme Courts' , The Social Science Journal 1995, Vol. 32 No. 4 Pages 441-457

Laband, D. N., and Lentz, B. F. (1985). 'Favorite sons: Intergenerational wealth transfers among politicians.' Economic Inquiry, 23, Pages 395-414.

Lancaster, TD and Patterson, DW (1990) 'Comparative Pork Barrel Politics: Perceptions from the West German Bundestag' Comparative Political StudiesVol 22 pages 458-77

Lawless, J. L. (2012).Becoming a Candidate: Political Ambition and the Decision to Run for Office. Cambridge University Press

Leeson, P (2008). "Media Freedom, Political Knowledge, and Participation."Journal of Economic Perspectives, Vol. 22(2): Pages 155-169.

Leenknegt GJ and van der Schyff G (2007) ‘Reforming the Electoral System of the Dutch Lower House of Parliament: An Unsuccessful Story’ German Law Journal Vol 8 No 12 available: https://static1.squarespace.com/static/56330ad3e4b0733dcc0c8495/t/56b86c99a3360cac2 760b18a/1454927019274/GLJ_Vol_08_No_12_Leenknegt.pdf [accessed 07/10/2017]

Levitt, SD Wolfram CD (1997) ‘Decomposing the Sources of Incumbency Advantage in the US House’ Legislative Studies Quarterly Vol. 22 No. 1 1997 available: http://www.jstor.org/stable/440290 [accessed 02/03/2018]

Lin, G and Rogerson P (1995) ‘Elderly Parents and the Geographic Availability of their Adult Children’ Research in Aging Volume: 17 issue: 3, page(s): 303-331 September 1, 1995 available: http://journals.sagepub.com/doi/abs/10.1177/0164027595173004 [accessed 08/03/2018]

Lott, JR (1986) “Brand names and barriers to entry in political markets” Public Choice 51 (1): pp. 87-92

Lott, JR (1987) ‘Political Cheating’ Public Choice Vol. 52 No. 2

220

Lott, JR (1989) “Explaining challengers’ campaign expenditure: The importance of sunk nontransferable brand name” Public Finance Quarterly 17(1): pp. 108-118

Lumb, M (2012) Parliamentary Relations: political families in the Commonwealth Parliament Parliament of Australia: Department of Parliamentary Services

Lumb, M and Goddett H (2015) Parliamentary Relations: political families in the Commonwealth Parliament Parliament of Australia: Department of Parliamentary Services

Komito, L (1984) ‘Irish Clientelism: A Reappraisal’ The Economic and Social Review Vol 15 No.3 April 1984 pp 173-194

Machiavelli, N (1503) Discourses on the First Decade of Titus Livy, Book I, Chapter 10 available: http://www.constitution.org/mac/disclivy1.htm#1:10 [accessed 15/06/2014]

McEnroe, J (2011) ‘Analysis: So how is the Government really doing/’ June 16th 2011 available: https://www.irishexaminer.com/text/ireland/kfojidaukfsn/ [accessed 19/03/2018]

McKean, M and Schneider, E (2000) 'Japan’s new electoral system: la plus c¸a change . . . ' Electoral Studies 19 (2000) Pages 447–477

Mair, P (1981) 'Analysis of Results: Dáil General Election 1981' Pages 150-154 in Nealon, T (1981) Nealon's Guide to the 22nd Dáil and Seanad Dublin: Platform Press

Mann, TE (2016) ‘We must address Gerrymandering’ Time October 13th 2016 available: http://time.com/4527291/2016-election-gerrymandering/ [accessed 10/09/2017]

Marsh, M (1987) Candidate Selection in Comparative Perspective: The Secret Garden of Politics London: Sage

Marsh, M (2007) 'Candidates or Parties: Objects of Electoral choice in Ireland' Party Politics Vol. 13. No.4 Pages 500–527

Martin, S (2011) 'Electoral Institutions, the Personal Vote and Legislative Organization' Legislative Studies Quarterly Vol. 36. 3 pages 339-361

Mehrotra, V Morck, R Shim, J and Yupana, W (2013) ‘Adoptive expectations: Rising sons in Japanese family firms’ Journal of Financial Economics June 2013 pp 840 -854

Mendoza, RU Beja, E Venida, V and Yap, D (2012) ‘ An Empirical Analysis of Political Dynasties in the 15th Philippine Congress’ Asian Institute of Management Policy Center Research Paper

Michels, R. (1962 [1911]), Political Parties. A Sociological Study of the Oligarchical Tendencies of Modern Democracy (London: Collier Macmillan).

Morck, R, Strangeland, D and Yeung, B (2000) 'Inherited Wealth, Corporate Control, and Economic Growth The Canadian Disease?' in Concentrated Corporate Ownership Pages 319- 372 CambridgeMA: National Bureau of Economic Research

221

Mosca, G (1896) La Classe Politica [first published as Elementi di Scienza Politica] (Bari: Laterza Editori). as quoted in Dal Bó, E, Dal Bó, P and Snyder J (2009) 'Political Dynasties' Review of Economic Studies (2009) 76, Pages 115–142

Nealon, T (1981) Nealon's Guide to the 22nd Dáil and Seanad Dublin: Platform Press

Nealon, T (1982) Nealon's Guide to the 23rd Dáil and Seanad Dublin: Platform Press

Nealon, T (1983) Nealon's Guide to the 24th Dáil and Seanad Dublin: Platform Press

Nealon, T (1987) Nealon's Guide to the 25th Dáil and Seanad Dublin: Platform Press

Nealon, T (1989) Nealon's Guide to the 26th Dáil and Seanad Dublin: Platform Press

Nealon, T (1993) Nealon's Guide to the 27th Dáil and Seanad Dublin: Gill and Macmillan

Nealon, T (1997) Nealon's Guide to the 28th Dáil and Seanad Dublin: Gill and Macmillan

New Zealand Electoral Commission (2017) available: http://www.elections.org.nz/home [accessed 20/01/2016-23/07/2017]

Norris, P (1997) ‘The Puzzle of Constituency Service’ The Journal of Legislative Studies Vol.3 No. 2 Pages 29-49

O’Riordan, S (2016) ‘A family affair as TDs pick siblings to ‘guard patch’ in Cork County Council’ Irish Examiner March 15 2016 available: https://www.irishexaminer.com/ireland/a- family-affair-as-tds-pick-siblings- to-guard-patch-in-cork-county-council-387419.html [accessed 19/03/2018] O’Rourke, M (2012) Just Mary:A Memoir Gill and Macmillan, Dublin Orsman, B (2000) ‘Key electorate: Coromandel’ New Zealand Herald available: http://www.nzherald.co.nz/politics/news/article.cfm?c_id=280andobjectid=16899 [accessed 16/07/2017]

Parliament of Australia (2013) The Role of the Senate available: http://www.aph.gov.au/About_Parliament/Senate/Powers_practice_n_procedures/briefs/b rief10#1 [accessed 15/9/13}

Parliament of Belgium (2016) De Kamerladen available: https://www.dekamer.be/kvvcr/showpage.cfm?section=/depute&language=nl&cfm=cvlist5 4.cfm?legis=53&today=n [accessed 18/02/2016]

Parliament of Canada (2015) Family Relations in the Canadian Houses of Parliament available: http://www.parl.gc.ca/parlinfo/compilations/parliament/FamilyTies.aspx [accessed 8/9/15]

Parliament of the Netherlands (2016) Het biografisch archief van PDC available: https://www.parlement.com/id/vhnnmt7ihwyl/biografisch_archief_pdc [accessed 23/08/2016]

Parliament of New Zealand (2016) MPs and Electorates available: https://www.parliament.nz/en/ [accessed 18/07/2016]

222

Patzelt, W (1997) ‘German MPs and Their Roles’ in Members of Parliament in Western Europe: Roles and Behaviour eds Müller, C Saalfeld,T (1997) Oregon: Frank Cass and Co. Ltd

Prewitt, K, Eulau, H and Zisk, B (1967) “Political Socialization and Political Roles,” Public Opinion Quarterly, 30 (1966-67): Page 573 as quoted in Kurtz, D (1995) 'Inheriting a political career: The justices of the United States and Louisiana Supreme Courts' , The Social Science Journal 1995, Vol. 32 No. 4 Page 449

Putnam, R (1976) The Comparative Study of Political Elites Prentice Hall: New Jersey

Querubin, P (2012) ‘Political Reform and Elite Persistence: Term Limits and Political Dynasties in the Philippine’s APSA 2012 Annual Meeting Paper

Rae, D W (1967).The Political Consequences of Electoral Laws New Haven and London: YaleUniversity Press

Recchi, E (1999) 'Politics as Occupational Choice: Youth Self-Selection for Party Careers in Italy' European Sociological Review, Vol. 15 No. 1, Pages 107-124

Reed, S (1994) 'Democracy and the Personal Vote: A Cautionary Tale from Japan.' Electoral Studies Vol.13 No.1 pages 17-28.

Renwick, A and Pilet, JB Faces on the Ballot: The Personalization of Electoral Systems in Europe Oxford 2016

Representation of the People Act (2002) Government of Norway available: https://www.regjeringen.no/globalassets/upload/KRD/Kampanjer/valgportal/Regelverk/Re presentation_of_the_People_Act170609.pdf [accessed 30/07/2017]

Riedinger, J (1995) 'The Philippines in 1994: Renewed Growth and Contested Reforms' Asian Survey, Vol. 35, No. 2, A Survey of Asia in 1994: Part II (Feb., 1995), Pages. 209-216

Rodrigues, M and Brenton, S (2010) The age of independence? Independents in Australian parliaments Parliament of Australia available: https://www.aph.gov.au/About_Parliament/Parliamentary_Departments/Parliamentary_Lib rary/pubs/rp/rp1011/11rp04 [accessed 07/04/2018]

Rossi, M (2009). ‘The causes of political dynasties in democratic countries’.Mimeo.available: http://economia. uniandes. edu. co/content/.../29957/.../Political_Dynasties_Rossi. pdf, 2009. [accessed 18/6/13]

Saks, P.M. (1976) The Donegal Mafia: An Irish Political Machine Yale: Yale University Press

Sartori, G (1976) Parties and Party Systems: a Framework for Analysis Cambridge University Press

Saisana, M and Saltelli, A (2012) Corruption Perceptions Index 2012 Statistical Assessment JRC Scientific and Policy Report, European Commission available: [accessed 23/07/17] http://www.chiletransparente.cl/wp- content/files_mf/1363104309JRC_Statistical_Assessment_CPI2012_FINAL.pdf

223

Sawer, M. (2004).‘Above-the-line Voting: How Democratic?’ source: http://democratic.audit.anu.edu.au/abovetheline.pdf [accessed 12/3/12]

Schatz, E (2004). “Modern Clan Politics: The power of ‘Blood’ in Kazakhstan and Beyond”. University of Washington Press

Schwarz, H (2014) ‘Voter turnout in primary elections this year has been abysmal’ Washinton Post July 23rd 2014 available: https://www.washingtonpost.com/blogs/govbeat/wp/2014/07/23/voter-turnout-in- primary-elections-this-year-has-been-abysmal/?utm_term=.ef3e5a8f0e5c [ accessed 27/03/2018]

Seawright, A and Gerring, J (2008) ‘Case Selection Techniques in Case Study Research: a Menu of Qualitative and Quantitative Options’ Political Research Quarterly Vol 61 No. 2 available: http://journals.sagepub.com/doi/abs/10.1177/1065912907313077 [accessed 23/05/2018]

Seeker, WP (1991) Ministers in beeld. Desociale en functioneleherkomst van de Nederlandse ministers (1848-1990).Leiden, DSWO Press Pages 102-105 as quoted in Van Liefferinge H andSteyvers K (2007) Does Family Matter? Socialization, Ambition and Career Development of Mayors in Belgium' accepted for ECPR – General Conference Pisa, 6-8 September 2007 Page 3.

Sheridan, K (2010) 'Putting Women in the Frame' The Irish Times available: http://www.irishtimes.com/indepth/oireachtas/putting-women-in-the-frame.htm [accessed 18/8/11]

Shugart, M (2001) ‘Electoral “efficiency” and the move to mixed member Systems’Electoral Studies Vol. 20 Pages 173-193

Sieberer, U (2010) ‘Behavioral consequences of mixed electoral systems: Deviating voting behavior of district and list MPs in the German Bundestag’ Electoral Studies Vol. 29 No. 3 Pages 484-496

Smith, DM (2012) ‘Succeeding in Politics: Dynasties in Democracies” UC San Diego Electronic Theses and Dissertations available: http://escholarship.org/uc/item/1dv7f7bb [accessed 13/06/2015

Smith, DM and Martin, S (2017) ‘Political Dynasties and the Selection of Cabinet Ministers’ Legislative Studies Quarterly Vol 42 Issue 1 February 2017 pp 131-165

Statistics Iceland (2010) Icelandic National Election Study 2009 available: http://fel.hi.is/2009_0 [accessed 07/10/2016]

Stephens-Davidowitz S (2015) ‘Just How Nepotistic Are We?’ New York Times March 21st 2015 available: http://www.nytimes.com/2015/03/22/opinion/sunday/seth-stephens- davidowitz-just-how-nepotistic-are-we.html?_r=0 [accessed 17/10/2015]

Steiner, N D (2012) ‘Testing for a Political Bias in Freedom House Democracy Scores: Are U.S. Friendly States Judged to Be More Democratic?’Journal of Comparative Policy AnalysisAvailable: http://www.academia.edu/5869279/Testing_for_a_Political_Bias_in_Freedom_House_Dem

224 ocracy_Scores_Are_U.S._Friendly_States_Judged_to_be_more_Democratic [accessed 15/8/14]

Sykes T (2013) ‘Iceland’s Incest-Prevention App Gets People to Bump Their Phones Before Bumping in Bed’ The Daily Beast available: http://www.theDailybeast.com/articles/2013/04/23/iceland-s-incest-prevention-app-gets- people-to-bump-their-phones-before-bumping-in-bed.html [accessed 23/04/2015]

Taoiseach (2013) 'Former Taoisigh’ Department of An Taoiseach available: http://www.taoiseach.gov.ie/eng/Historical_Information/Former_Taoisigh/ [accessed 02/12/13]

Terry, C (2012) ‘Serving up a Dog’s Breakfast’ Electoral Reform Society available:http://www.electoral-reform.org.uk/blog/serving-up-a-dogs-breakfast [accessed 15/9/13]

Transparency (2014) Corruption Perception Index Transparency International available: http://www.transparency.org/cpi2014/ [accessed 21/11/2015]

Transparency (2016) Corruption Perception Index Transparency International available: https://www.transparency.org/news/feature/corruption_perceptions_index_2016#table [accessed 21/06/2017]

United Nations (2013) ‘Human Development Report 2013:The Rise of the South Human Progress in a Diverse World’ United Nations Development Programme available: http://hdr.undp.org/en/data/profiles [accessed 9/9/13]

US Congress (2016) Biographical Directory of the United States Congress 1774 – Present available: http://bioguide.congress.gov/biosearch/biosearch.asp [accessed 07/06/2016]

Uslaner, E (1985) ‘Casework and Institutional Design: Redeeming Promises in the Promised Land’ Legislative Studies Quarterly, Vol. 10, No. 1 Pages 35-52

Van Coppenolle, B (2014) Political dynasties and elections. PhD thesis, The London School of Economics and Political Science (LSE). Available: http://etheses.lse.ac.uk/883/1/__lse.ac.uk_storage_LIBRARY_Secondary_libfile_shared_rep ository_Etheses_Content_Theses%20submitted%20by%20students%20%26%20alumni_Live %20theses_Van%20Coppenolle_Political%20dynasties_2014.pdf [accessed 06/12/2015]

Van Liefferinge, H and Steyvers, K (2007) Does Family Matter? Socialization, Ambition and Career Development of Mayors in Belgium' accepted for ECPR – General Conference Pisa, 6- 8 September 2007

Warren, DE and Laufer, WS (2010) ‘Are Corruption Indices a Self-Fulfilling Prophecy? A Social Labeling Perspective of Corrupyion’ Journal of Business Ethics October 2009, Volume 88, Supplement 4, pp 841-849 available: https://link.springer.com/article/10.1007/s10551-009- 0316-5 [accessed 20/07/2017]

Wasson, E A (1991) 'The House of Commons, 1660-1945: Parliamentary Families and the Political Elite' The English Historical Review, Vol. 106, No. 420 (Jul., 1991), Pages 635-651

225

Wood, DM Young, G (1997) ‘Comparing Constituency Activity by Junior Legislators in Great Britain and Ireland’ Legislative Studies Quarterly Vol 22 No. 2 available: http://www.jstor.org/stable/440383 [accessed 23/01/2012]

226